A neuro-advertising property video recommendation system

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Technological Forecasting & Social Change journal homepage: www.elsevier.com/locate/techfore

A neuro-advertising property video recommendation system A. Kaklauskasa,⁎, E.K. Zavadskasa, A. Banaitisa, I. Meidute-Kavaliauskienea, A. Libermanb, S. Dzitacc, I. Ubartea, A. Binkytea, J. Cerkauskasa, A. Kuzminskea, A. Naumcika a b c

Vilnius Gediminas Technical University, Vilnius, Lithuania Nemesysco, Kadima, Israel Aurel Vlaicu University of Arad, Arad, Romania

A R T I C L E I N F O

A B S T R A C T

Keywords: MCDA Property Methods Neuro decision matrix Affective computing Knowledge-based real-world applications

Many factors influence the identification of the best real estate alternatives, such as supply and demand and the social, cultural, psychological and personal factors affecting buyer behavior and the emotional state of a buyer. What are the most effective ways of choosing a property, when the selection is so vast and complex? An aid is developed here to accomplish this, based on a new advertising format, an iterative method and the NEuroAdvertising Property Video Recommendation System (NEAR). A known methodology involves behavioral operational research and the emotions involved in decision-making. Three advanced research contributions are unique to the proposed method and NEAR, in contrast to innovative behavioral operational research. Firstly, data are compiled for a neuro decision matrix, based on housing attributes and the valence, arousal, emotional state and physiological parameters of a potential real estate buyer. Secondly, the performance of a multiple criteria neuroanalysis occurs as well as the selection of the most personalized and effective video clip ad variants drawn from many alternatives. Finally, NEAR is found to present the most effective video clips ads for real estate buyers for as long a period as possible, according to Multiple Resource Theory.

1. Introduction A founder of behavioral economics, Nobel Prize laureate Kahneman (2011) asserts that two categories describe our thinking: fast thinking (first system) and slow thinking (second system). The foundation of the first system consists of emotions, impulses and exaggerated optimism. This system does not require any great effort; it operates almost automatically. In contrast, the second thinking system is slow and analytical, and has the ability to control behavior and thoughts. Hämäläinen et al. (2013) draw attention to the need for behavioral operational research (BOR) to support human problem-solving by modeling the practice of OR in advance. These authors show that completely opposite results can be derived depending on the way the phenomenon is described, how the questions are phrased and which graphs are used; their results suggest that OR processes are highly sensitive to various behavioral effects. A growing stream of research has examined emotions and decision-making based on the appraisal tendencies associated with emotions (So et al., 2015). Initial theories described decision making as a psychophysical process; modern approaches, however, take into account factors such as emotion and intuition (Chick et al., 2017). Nobel Laureate Simon (1997) analyzes the role of emotions in

⁎

decision-making and concludes that there is no intrinsic conflict between rationality and emotion, and that emotion can be conducive to making good decisions. Mikels et al. (2015) provide an overview of theoretical perspectives that emphasize the role of emotion in decisionmaking. Koshkaki and Solhi (2016) provide empirical evidence that negative emotions (fear, anger and shame) significantly facilitate decision making. Treur and Umair (2015) analyze emotions as a vehicle for rationality. They investigate rational decision-making models based on emotion-related valuing. People often shop when feeling sad (Rick et al., 2014). Garg and Lerner (2013) have conducted research showing that sadness influences consumption, leading individuals to pay more to acquire new goods and to eat more unhealthy food than they would otherwise. Some researchers (Bastiaansen et al., 2016; Cash et al., 2017; Challcharoenwattana and Pharino, 2016; Gohary and Hanzaee, 2014; Qin et al., 2017; Wang et al., 2013) have also applied behavioral multiple criteria decision making in their studies. Globally, there are developments of multimodal (Lee and Norman, 2016; Esposito et al., 2015; Poria et al., 2017; Poria et al., 2016; Dai et al., 2015; Gauba et al., 2017; Chen et al., 2016; Mehmood et al., 2016; Ringeval et al., 2015) video biometric and affective computing systems, which endeavor to analyze user emotions.

Corresponding author. E-mail address: [email protected] (A. Kaklauskas).

http://dx.doi.org/10.1016/j.techfore.2017.07.011 Received 6 January 2017; Received in revised form 14 June 2017; Accepted 11 July 2017 0040-1625/ © 2017 Elsevier Inc. All rights reserved.

Please cite this article as: Kaklauskas, A., Technological Forecasting & Social Change (2017), http://dx.doi.org/10.1016/j.techfore.2017.07.011

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

expressions are monitored while a video ad is playing. This facilitates better control over alternative video clip ads, or even allows the process to be halted altogether, if no suitable video clip ads can be found in the property video clip ads database for the respective viewer. NEAR analyzes, in real time, the buyer's feelings: anger, surprise, happiness, disgust, sadness, fear and neutrality. This analysis of the viewer's emotions helps in presenting that viewer with those video clip ads which he or she would prefer at that point. Following this, the model for evaluating a user's interest in a specific video clip makes a real-time selection of the most effective property ad. It performs this using an integrated assessment of the viewer's completed personalized questionnaire, video clip content, valence, arousal, emotional expressions on the potential buyer's face (happy, sad, surprised, scared, disgusted or neutral) and his or her physiological parameters.

On a daily basis, the Internet offers increasing amounts of ever more varied data and information regarding widely differing real estate objects for sale. Potential real estate buyers are highly heterogeneous, differing from one another in terms of age, sex, level of education, income, marital status, social standing, psychological state, lifestyle and emotional and cultural features, all of which influence their needs. Differences in feelings, motivations, personalities, temperaments and moods are often intertwined with the emotions of potential real estate buyers. All of these data and information frequently originate as diverse modalities with dissimilar statistical properties. The processing and analysis of all these multimodal data, along with interpretations of the emotional states of potential real estate buyers followed by the submission of information most pertinent to them, are extremely complicated; however, these are of utmost importance to the real estate and construction industries. A multiple criteria, multimodal, neuroanalysis framework must therefore be developed for this purpose. This paper aims to analyze the effectiveness of the real estate advertising process in terms of increasing interest by applying intelligent and multimodal biometric technologies. These technologies serve as the basis for developing the Neuro-Advertising Property Video Recommendation System (hereafter, NEAR). NEAR is intended to analyze the real estate objects under deliberation, the needs and emotions of potential real estate buyers and their interest in pieces of advertising. It is also intended to rationalize the advertising process, based on the best worldwide practices and the current states of the audience. The organization of this manuscript is as follows: Sections 2 and 3, which follow this introduction, describe the NEAR method and system. Section 4 presents case studies, and concluding remarks appear in Section 5.

2.2. Compilation of the neuro decision matrix One of the most important stages of a multiple criteria decision analysis involves the establishment of a system of criteria describing the alternatives, measurement units, weights and values. The video clip ads to be shown (from the Property Video Clip Ads Database) relate directly to the quantitative and qualitative data of these alternatives (Real Estate Database) comprehensively describing the alternatives under consideration. The compilation of data for the neuro decision matrix is based on the Real Estate Database and the emotional state determined for each potential buyer while reviewing alternatives. This data comprehensively describe the real estate alternatives for that buyer. NEAR captures criteria together with the information which describes them (measuring units of the criteria, values [x11 − xtn] and weights [q1 − qt]) from the Real Estate Database. It also captures criteria Xt + 1 − Xt + 7 from the Microsoft Emotion API Affective Database (measuring units of the criteria [mt + 1 − mt + 7], values [xt + 1 1 − xt + 7 n] and weights [qt + 1 − qt + 7]) and criterion Xt + 8 from the QA5 SDK Emotions Database (measuring unit of the criterion [mt + 8], values [xt + 8 1 − xt + 8 n] and weight [qt + 8]). This completes the compilation of a neuro decision matrix for a specific viewing buyer (see Table 1). The elements of a neuro decision matrix indicate solutions based on specific behavioral operational research and emotions in decision making. The neuro decision matrix is valuable in assessing a system of decision-making aspects and evaluating the comparative importance of each aspect. A multiple criteria real estate analysis problem can be characterized by a neuro decision matrix. The methods used to establish the degree of utility and priorities of the variants under comparison are calculations of criteria values and weights and the application of multiple criteria analysis for projects.

2. The NEAR method 2.1. Establishing the most effective video clip ads Studies of body language suggest that in face-to-face communication, silent signals account for 60%–80% of the speaker's impact on the person spoken to, voice accounts for 20%–30%, and words account for the remaining 7%–10%. Real estate brokers can observe and analyze a viewer's emotional state during commercials (by reading facial expressions and body movements, understanding voice inflections such as tone or intonations and the like) thereby becoming more efficient in communicating with a potential real estate buyer and selecting a more rational advertising style. When it comes to interaction (feedback) with potential buyers, real estate brokers beat current neuroadvertising systems (NAS) in terms of efficiency. One main reason for this is that a real estate agent can read a buyer's body language, which is a means of effective communication that helps to better understand buyers. Focusing attention on the buyer's words and scrutiny of the buyer's gestures or movements can disclose the inner states and emotions experienced while the person is exposed to real estate ads. The NEuro-Advertising Property Video Recommendation System (NEAR) analyzes the requirements of potential buyers for real estate (with the aim of establishing the most appropriate video advertisement and performing a multiple criteria decision analysis on housing, with the goal of selecting the most effective variant). It also recognizes, interprets, processes and analyzes the buyer's emotions and interest in video clip ads, and proposes the most effective guidelines to make these video clip ads more interesting. This analysis of ad rationality takes into account the viewer's completed Personalized Questionnaire, video clip content (the opinion analytics of video texts), valence, arousal, emotional state (the Microsoft Emotion API (the integrated Microsoft Emotion API with FaceReader 5.0) Subsystem) and physiological parameters. This allows optimization of the various permutations of the set of video clips. As a potential buyer watches a video ad, NEAR discovers further details of that person's emotions and responses. The viewer's facial

2.3. The NEAR method The integration involved in construction of the NEAR method includes self-analysis, Maslow's hierarchy of needs theory, four multiple criteria decision analysis methods developed by the applicants, and biometric, statistical (LOGIT, KNN, MBP, Rprop), recommender, big data and text analytics methods. The purpose of this research is to develop a new method for the property video neuroadvertising field. The proposed method matches the needs of potential property buyers with the existing housing market, based on multi-variant design and multiple criteria decision analysis procedures, a criteria system and a weighted tree structure applied to the selection of the most appropriate video clip content. Even if potential buyers have only a vague idea of the kind of property they are looking for, NEAR can offer valuable assistance in finding the most fitting property by analyzing a completed Personalized Questionnaire, video clip contents, valence, and a user's arousal, emotional state and physiological parameters. Reasoning and electronic negotiations form the basis of the NEAR model for 2

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

Table 1 Neuro decision matrix. Quantitative and qualitative information for alternatives Criteria describing the alternatives

Real estate database X1 X2 ... Xi ... Xt

a

l1 l2 ... li ... lt

Weight

Measuring units

q1 q2 ... qi ... qt

m1 m2 ... mi ... mt

Real estate alternatives under comparison 1

2

…

j

…

n

x11 x21 ... xi1 ... xt1

x12 x22 ... xi2 ... xt2

… … … … … …

x1j x2j ... xij ... xtj

… … … … … …

x1n x2n ... xin ... xtn

… … … … … …

xt + 1 xt + 2 xt + 3 xt + 4 xt + 5 xt + 6 xt + 7 xt + 8

Data for emotional state of the viewer when analyzing alternatives (Microsoft Emotion API Affective Database and QA5 SDK Emotions Database) Neutral, Xt + 1 lt + 1 qt + 1 mt + 1 xt + 1 1 xt + 1 2 … Happy, Xt + 2 lt + 2 qt + 2 mt + 2 xt + 2 1 xt + 2 2 … Sad, Xt + 3 lt + 3 qt + 3 mt + 3 xt + 3 1 xt + 3 2 … Angry, Xt + 4 lt + 4 qt + 4 mt + 4 xt + 4 1 xt + 4 2 … Surprised, Xt + 5 lt + 5 qt + 5 mt + 5 xt + 5 1 xt + 5 2 … Scared, Xt + 6 lt + 6 qt + 6 mt + 6 xt + 6 1 xt + 6 2 … Disgusted, Xt + 7 lt + 7 qt + 7 mt + 7 xt + 7 1 xt + 7 2 Evaluation of emotional states with QA5 SDK Subsystem, Xt + 8 lt + 8 qt + 8 mt + 8 xt + 8 1 xt + 8 2

xt + 1 xt + 2 xt + 3 xt + 4 xt + 5 xt + 6 xt + 7 xt + 8

Property video clip ads database Vk

Vj

a

V1

V2

…

j j j j j j j j

…

n n n n n n n n

Vn

The sign + (−) indicates that a greater (lesser) criterion value corresponds to a greater (lesser) significance for stakeholders.

state of a potential real estate buyer. This subsystem consists of the full set of equipment required by NEAR to analyze a viewer's nonverbal information and to assess the neurobiological response to the video clip ads. This equipment includes the integrated Microsoft Emotion API with FaceReader 5.0, a video camera, the QA5 SDK, a Mirametrix S2 Eye-Tracker, a Flir Thermo Cam B2, an Enobio Helmet, a voice stress analysis subsystem, a monitoring system Omron InteliSence M2-Basic, an Extech MO270 with a wireless pulse oximeter, a wireless body thermometer, a wireless smart gluco-monitoring system and a Polar h3 heart sensor. The equipment subsystem allows the acquisition of a great deal of multimodal data. These data are stored in the database and used in the calculations for the model-base. In its current version, NEAR can sort alternative video ads for property for sale into clusters, based on answers to the Personalized Questionnaire on environmental influences, individual differences and housing attributes. The dimensions of a buyer's social standing (education and income), family and household influences (age, marital status, number of children and family size) appear as additions to the Personalized Questionnaire to consider environmental factors. Quester et al. (2011) believe that the distinctions between social classes cause differences in patterns of consumption behavior and noticeably impact the “evaluation of alternatives” stage of buyer purchase decision making. The effect of family and household influences is similar. This article aims to showcase the potential of the proposed NEAR. Inclusion of environmental factors such as culture, ethnicity, group and personal influences in the Personalized Questionnaire would involve a wide scope of research; for example, if culture were to appear as a factor, it would then be necessary to analyze the values, behaviors, beliefs, knowledge, ideas and other traits of the buyer as a member of society. An analysis of group and personal influences would require a focus on a potential property buyer's mutual relationships with fellow employees, friends, family members and other individuals. These factors were therefore excluded from the development of NEAR in order to limit the scope of this research, even if culture, ethnicity, group and personal influences have been shown to affect the needs and behavior of a potential property buyer. The incorporation of these factors into NEAR is left for future work. NEAR sorts alternative videos from the Property Video Clips Ads Database into clusters, based on the viewer's answers to the

determining the most rational real estate purchase alternative and model for negotiations based on real calculations. NEAR can also help improve ads. One way of explaining a purchase made by a real estate buyer relates to an understanding of the needs that buyer is attempting to satisfy and an examination of the factors that inspire buyers to make a specific purchase. The theory developed by Maslow (1943) is an important method for understanding buyer needs. According to this theory, there are five levels of needs determining human behavior (Maslow, 1943): physiological needs, safety needs (dangers, threats and hardship protections), social needs (group membership), esteem and recognition needs (reputation, status, respect and recognition from others), and selfactualization needs (self-expression regarding one's own potential involving continuous improvements). Two projects (Kaklauskas et al., 2015a, 2015b) have included studies which aim to determine “the dependence of multimodal biometric parameters of office workers on stress” and “the dependence of the productivity of office workers on multimodal biometric temperature data”. Analysis of these dependencies involved the use of LOGIT, knearest neighbors (KNN), Marquardt-Back propagation (MBP) and resilient back propagation (Rprop) algorithms. These methods are widely used in biometric data analysis in studies worldwide. However, whenever biometric data is analyzed, different methods may produce different results; these studies therefore use combinations of several appropriate methods. The development of the NEAR method was therefore based on an integration of the LOGIT, KNN, Marquardt-Back propagation (MBP) and resilient back propagation (Rprop) methods. 3. The NEAR architecture and system The basis for the development of the NEuro-Advertising Property Video Recommendation System (NEAR) is the NEAR method. The following subsystems comprise NEAR: an equipment subsystem, a database and its management subsystem, a model-base and its management subsystem and a user interface (see Fig. 1). A brief analysis of these components is presented below. 3.1. The equipment subsystem and the personalized questionnaire The equipment subsystem aims to effectively assess the emotional 3

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

Fig. 1. The NEAR architecture. Fig. 2. Personalized Questionnaire identifying environmental influences, buyer's individual differences and housing attributes.

alternative video ads of properties for sale. The Dependencies Database discovers the dependencies linking the Personalized Questionnaire with the Property Video Clips Ads Database. An analysis of scientific articles and statistical data (Hilton Head Real Estate Partners, 2016; Kvietkutė, 2013; McLaughlin, 2015; Morgan and Media, 2016; NAR, 2015; National Association of Realtors, 2016; Pant, 2012; Reeves, 2015; Romero and Bezdeka, 2014; Trapasso, 2016; Wilson and Wilson, 2015), conducted by the current authors shows that age, gender, education, marital status, type of deal and other factors all affect buyer needs and buying habits. Buyers aged 18–34, for instance, typically seek cheaper properties and commit smaller sums for buying property than older buyers, due to their smaller incomes. A look

Personalized Questionnaire (see Fig. 2) on environmental influences, individual differences and housing attributes. Each cluster of property video ads includes a selection of video ads based on these factors of environmental influences, individual differences and housing attributes; identical responses to a video clip are expected from buyers within each segment. NEAR is a massive system designed to satisfy the video ad needs of individual property buyers. Although alternative property video ads are currently sorted based on environmental influences, individual differences and housing attributes, future work may involve the addition of a feature allowing users to choose for themselves which environmental influences, individual differences and housing attributes should be considered in the sorting of 4

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

Fig. 3. Excerpt from Dependencies Database linking questions from the Personalized Questionnaire with property video clip ads.

• The Microsoft Emotion API Affective Database • The Microsoft Emotion API Valence and Arousal Database • The QA5 SDK Emotions Database • The Real Estate Database • The e-Negotiation Database • An intelligent database engine

at the differences in property choices between the genders shows that women tend to focus on the sizes of the kitchen, bathroom and wardrobes, whereas men are more interested in recreational spaces, pools and tubs. Marital status may also shed light on the size of and requirements for real estate investments. Wilson and Wilson (2015) contrast married and single buyers, finding that married buyers have larger incomes and can afford better housing. These buyers place emphasis on an environment for their children to grow in and the affordability of housing. The case of single buyers, also differentiated by gender, is quite different: single men prefer properties closer to entertainment, sports and recreation centers, whereas single women choose properties which are closer to their family or friends. Fig. 3 shows an excerpt from the Dependencies Database, which links questions from the Personalized Questionnaire with property video clip ads. Once these dependencies have been established, the buyer is shown property video clip ads best matched to his or her needs. It is important for real estate brokers to understand the reactions of buyers when exposed to marketing stimuli such as needs, motives, personality, experiences and expectations. Individuals often identify the same kinds of stimuli in different ways, as they often focus on elements which are significant to their needs (Belch and Belch, 2012). When a potential property buyer completes the Personalized Questionnaire, the number of possible video ads for that buyer is reduced; in theory, only ads which will interest the buyer personally will remain. In practice, however, only a fraction of the remaining ads may appeal to the buyer. For this reason, the Microsoft Emotion API Subsystem is used to analyze a viewer's emotions and only displays the ads which the user likes most.

A brief analysis of these databases is given below. The Property Video Clip Ads Database comprises various videos promoting properties for sale. The compiled Personalized Questionnaire ensures that a buyer is only offered those properties that meet his or her needs (see the excerpt in Fig. 3). The Personalized Questionnaire determines a buyer's age group, gender, education, income, marital status, type of property deal and related aspects. Data analytics is used to search the Personalized Questionnaire and Answers Database, along with the other databases, for various dependencies and trends. Any trends identified are included in correlation tables, which show the links between the user's emotions and physiological parameters. The Correlation Database discovers the dependencies between the user's emotions, valence, arousal and physiological parameters (heart rate, blood pressure, pupil size, skin conductance and humidity, body temperature) and video clip ads aj (priority Pj and degree of utility Nj). The text database of video clips contains all the text from the video clips in English, Russian or Lithuanian and a list of keywords for indexing each video text. The historical statistics database accumulates historical statistical data: statistical analysis of the responses of users, statistical analysis of the video clips watched, and the predominant emotions, valence and arousal states displayed while the users watched the video clips. Additional accumulated information includes how many times the viewing of a video was terminated and the places where it was terminated due to insufficient relevance and interest to the viewer, and how many times viewers watched each video to completion. Conclusions can be drawn based on this information regarding places in the advertisement which require improvement. Data obtained from analyzing a user's physiological parameters with the help of the equipment subsystem are entered into that user's Physiological Parameters Database.

3.2. The database and its management subsystem The NEAR database comprises the following:

• The Property Video Clip Ads Database • The Personalized Questionnaire and Answers Database • The Dependencies Database • The Correlation Database • A text database of video clips • A historical statistics database • The User's Physiological Parameters Database 5

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

1. Non-versioned data, including the Property Video Clips Ads Database, Text Database of Video Clips, Real Estate Database and eNegotiation Database (as the needs of the buyer change, the data becomes outdated or inaccurate; older versions are then no longer of value and are simply removed from the databases); 2. Versioned data generated at regular intervals, including the Microsoft Emotion API Affective Database, Microsoft Emotion API Valence and Arousal Database and the User's Physiological Parameters Database (different data versions from different periods are necessary and are therefore stored; a version number and date are essential for these collections of data).

Data obtained from analyzing a user's face using the Microsoft Emotion API (the integrated Microsoft Emotion API with FaceReader 5.0) Subsystem are entered into the Microsoft Emotion API Affective, Valence and Arousal databases. Users specify certain requirements and constraints, and NEAR then requests information on specific real estate from a number of online brokers. Search results for specific real estate appear in the Real Estate Database (textual, photo/video and graphical information on the real estate alternatives) and the e-Negotiation Database. The Intelligent Database Engine consists of two main parts: 1) text mining; and 2) determination of the interdependencies between the interest of the potential real estate buyer under analysis and his or her physiological parameters. All the data in the NEAR database are stored in tables and organized using relational database principles. These are then used in a typical relational Intelligent Database Management Subsystem. The most important functions of the Intelligent Database Management Subsystem are the design of the structure of a database; loading, accumulating and editing a database; reviewing, searching, sorting and other arrangement of the data; creating user application programs and compiling reports; and application of the Intelligent Database Engine. The collection of all the required data is insufficient to ensure the optimal use of the NEAR database. The data must be updated over time, requiring efforts to limit excessive, unnecessary information. In the following, we give an overview of the NEAR databases, classified in terms of their storage periods:

• Data which is not stored includes: 1. Temporary data such as the interim results of various calculations; 2. Calculated (derived) data generated as required from the main data stored in a database by querying and pre-programming tasks (age data is not stored, since dates of birth constitute main, fixed and permanently stored data, and age is therefore easy to calculate). 3.3. The model-base and its management subsystem The following models comprise the model-base: the Multimodal Physiological Subsystem; the QA5 SDK Subsystem; the Microsoft Emotion API Subsystem; the model for evaluating user interest in a specific video clip; data analytics; opinion analytics of video texts; the model used to determine the most rational real estate purchase alternative; and the model for negotiations based on real calculations. Descriptions of these applied models and subsystems are given below. Firstly, the Multimodal Physiological Subsystem and the Microsoft Emotion API Subsystem assess integrated neurobiological viewer responses (physiological parameters, emotional states, valence and arousal). Then the model which evaluates user interest in a specific video clip decides on the most efficient property ad selection in real time using an integrated assessment of valence, arousal and the emotions shown on the buyer's face (happiness, sadness, surprise, fear, disgust and neutrality) and by the user's physiological parameters. The model which evaluates user interest in a specific video clip then decides whether to show the selected video clip ad. This decision is made under the following circumstances:

• NEAR databases with permanent storage: 1. Plain, permanent information is stored unaltered during the entire time that a specific dataset exists. These data are never altered but are repeatedly reused for various purposes. These form the standardized, historical data of the NEAR. When the volume of data (the Personalised Questionnaire and Answers, Dependencies and Correlation Databases) exceeds the storage capacity, these are archived. 2. Archived data are also permanently stored. These data may be required in the future, and thus are inherently important. However, they are complex, take up a relatively large space and in practice, their use is rare. Data archiving can cut data storage costs and streamline stored information management.

• The Microsoft Emotion API Subsystem uses a webcam to detect a

• Two types of updatable information are used in NEAR:

viewer's emotions (happiness, sadness, surprise, fear, disgust and neutrality) every two seconds (see Fig. 4). If the overall emotional

Fig. 4. Assessing the integrated neurobiological response of a potential real estate buyer based on facial emotions, valence and arousal.

6

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

Valence Unpleasant

Property involvement

Pleasant

Activated

Fig. 5. NEAR provides a way to present the most effective video clips ads for real estate buyers for as long a period as possible, according to Multiple Resource Theory.

Most effective video clip ads for buyers who are not

Arousal

concerned with a property (upper right part of the

BNCP

Unpleasant

Pleasant

Valence BGCP

circumplex model of emotion) Most effective video clip ads for buyers who are greatly concerned with a property (lower right part of the

Deactivated

circumplex model of emotion)

•

•

if the subsystem determines high (activated) arousal in a GCP viewer (or medium-low/medium-deactivated arousal in an NCP viewer), it suggests moving to the next video ad (see Fig. 5). Meanwhile, if the red dot for an NCP viewer (indicating the level of arousal) falls into the upper right-hand area of the circumplex model of emotion or into the lower right-hand area for a GCP viewer, then the subsystem recommends showing the current video ad. When the opposite occurs, then it recommends not showing it (see Figs. 4–6).

state detected by the subsystem is negative, it suggests skipping to the next video ad; however, if the overall emotional state is neutral/ positive, the subsystem suggests continuing with the current video ad. The Microsoft Emotion API Subsystem detects a viewer's valence level (see Fig. 4). The valence indicates whether the viewer's emotional state is positive (pleasant), neutral or negative (unpleasant). If the subsystem identifies a pleasant valence in the viewer watching an ad, it suggests continuing with the current video ad; if the valence detected by the subsystem is unpleasant or neutral, it suggests skipping to the next video ad. In other words, whenever the red dot falls on the right-hand side of the circumplex model of emotion, the subsystem recommends showing the current video ad; if not, then skipping it is recommended (see Figs. 4-6). The Microsoft Emotion API Subsystem detects a viewer's arousal level. The Personalized Questionnaire filled out by a potential buyer notes the buyer's viewpoint, that is whether or not he/she is greatly concerned with a property (GCP). NEAR generates a medium-deactivated level for a buyer who has noted being GCP in his/her Personalized Questionnaire. However, if a viewer has noted as being not concerned with a property (NCP), then an activated level is created. If the subsystem determines high (activated) arousal in an NCP viewer (or medium-low/medium-deactivated arousal in a GCP viewer), it suggests continuing with the current video ad. However,

NEAR only shows selected video ad clips when the emotions, valence and arousal of a potential property buyer confirm that the viewer likes the ad. For much of the last century, the idea guiding the psychology of emotion perception is that emotions are written on the face as particular arrangements of facial actions and that perceivers can read these actions as easily and effortlessly as they read words on a page (Barrett et al., 2011). Dahl and Gordon-Wilson (2013) suggest that campaigns perform better when they target consumers of advertising who are interested in the product. Belanche et al. (2017) find that high-arousal stimuli meet viewer expectations regarding relevant content, which increases the effectiveness of advertising. The positive or negative valence established on the face of a potential buyer reflects that buyer's view of the attractiveness (positive valence) or averseness (negative valence) of the real estate. Thus, NEAR helps to establish the valence on

Fig. 6. Change in a property ad, when the integrated assessment of facial emotions, valence and arousal indicates its unimportance with regard to the viewer's needs.

7

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

price, signing contracts, arranging for inspections, and assisting with closing. Conditions in the real estate market typically change rapidly, and there is both a large supply and demand for real estate. Thus, it is difficult to make rational decisions without the help of databases and intelligent systems and knowledge (Kaklauskas et al., 2012; Kanapeckiene et al., 2011). A model is developed here for determining the most rational real estate purchase alternative using neuro decision matrices (see Table 1) and multiple criteria methods (Kaklauskas, 1999, 2016) with the aim of determining the priority, degree of utility and market value of the real estate alternatives under analysis. A potential buyer performing a multicriteria analysis of all real estate alternatives selects several objects for initial negotiation. For that purpose, he or she marks the desirable negotiation objects (by ticking a box with a mouse). These calculations are performed based on characteristics describing the real estate alternatives obtained during negotiations (the neuro decision matrices). Based on the results received, the final neuro decision matrix is then developed. Following the development of the final neuro decision matrix, the multiple criteria decision analysis and selection of the best real estate buying version are carried out by applying the model to determine the most appropriate real estate purchase alternative. Buyers in the property market are often faced with an enormous supply and range of properties, thus hindering the straightforward choice of a property. One example involves housing attributes (Alkay, 2015; Ferreira and Jalali, 2015; Martins et al., 2015); researchers (Alkan, 2015; Deaconu et al., 2016; Jayantha and Lau, 2016) commonly sort housing attributes into intrinsic (housing type, layout, age, size, price, plot size), extrinsic (exterior housing design such as the aesthetics of the house, quality of the roof and external walls), space (garden and public area), environmental factors (pollution, noise, open space, parks) and location of facilities and services (public transport, schools, security, shopping centers, sports facilities). Hawkins et al. (2011) state that the evaluative criteria are typically the product features or attributes that buyers associate with the benefits they desire. As reported by Hawkins et al. (2011), there are a very large number of options available to buyers; however, buyers cannot be aware of all the available options for making their purchase decisions. The model developed here to determine the most rational real estate purchase alternative and the model for negotiations based on real calculations therefore help by reducing the information overload and improving both the accuracy of the buyer's decision and their satisfaction with it. This model-base provides the user with the opportunity to use the wide spectrum of the above models to solve specific problems.

a viewer's face. If this is a negative or neutral valence, NEAR suggests skipping to the next video ad; if it is a positive valence, then NEAR suggests continuing with the current video ad (see Fig. 4). Various authors have expressed the opinion that product involvement relates to the interests, values and needs of a potential buyer. In the opinion of Gunter et al. (2002), high-arousal stimuli may not be necessary for viewers who are highly involved with a product, since they are already interested in the advertisement. Other analogous studies (Bakalash and Riemer, 2013; Belanche et al., 2017) show that if a potential buyer is greatly concerned with a product, then it is appropriate to consider a medium-deactivated level as this buyer's arousal level, whereas if a viewer is not concerned with a product, then it would be effective to maintain the arousal level of that buyer at an activated level. Bakalash and Riemer (2013) hold the opinion that when an ad is less arousing for a viewer, it tends to be processed poorly, due to the lack of salience of any stimuli. The Personalized Questionnaire filled out by a potential buyer identifies the buyer's viewpoint, that is, whether or not he/she is greatly concerned with a property (GCP). NEAR maintains a medium-deactivated level for a buyer who is noted as being GCP in his/her Personalized Questionnaire. However, if a viewer is noted as not being concerned with a property, then it maintains an activated level for him/her. Figs. 4 and 6 show examples of the analysis of facial emotions, valence and arousal. NEAR analyzes viewer interests continually while examining the answers from the viewer's Personalized Questionnaire, the video clip content (Opinion Analytics of Video Texts), emotions (Microsoft Emotion API Subsystem) and physiological parameters. It then suggests a better video ad based on this rich nonverbal information. When a property ad contains information the viewer finds relevant and interesting, the NEAR continues with the same ad. However, the system skips to the next property ad when the integrated assessment shown above indicates that the current property ad is unimportant for the viewer's needs. Data analytics determines the correlations between the interest (emotions, valence and arousal) shown by the potential real estate buyer under analysis and his or her physiological parameters (heart rate [see Fig. 7], blood pressure, pupil size, skin conductance and humidity, body temperature) and the video clips ads. Data analytics is also used to perform statistical analysis of the questions selected by users, the video clips watched, and the predominant emotions, valence and arousal states displayed while watching the video clips. Furthermore, data analytics is used to accumulate information on how many times a shown video was switched off due to its insufficient relevance and interest to the viewer and at which point this happened, as well as how many times a video was watched through to its conclusion. This information can serve as the basis for deciding the points within an ad that require improvement. The Multimodal Physiological Subsystem processes data from the User's Physiological Parameters Database and uses the LOGIT technique (a regression model for ordinal dependent variables) to identify the potential real estate buyer's interest in the video clip ads. The correlations are then established between the interest shown in the video clip ads by the potential real estate buyers and their biometrical parameters. The correlation analysis of the interest of potential real estate buyers and his/her heart rate are based on the Logit technique, as shown in Fig. 7a. Dependency is established on the basis of 375 measures. Recognition levels of 41.1% and 84% with one point of error were established using a scale for the interest in video clip ads (see Fig. 7b).

4. Case study: compiling a neuro decision matrix of land lot alternatives and decision-making 4.1. An analysis of five land lot alternatives in Vilnius City The analysis in this case study involves five land lots in Vilnius City offered for sale by Capital PRO (see Table 2): 1. Saltoniškių Street Lot is land designated for commercial use in the Žvėrynas neighborhood. There are no restrictions due to city networks or cultural heritage objects, and all utility lines have been installed. The area of this lot measures 9.4 ares with an additional 0.97 ares available for purchase from the government. 2. Kęstučio Street Lot is land designated for investment. The use of this lot involves a residential site for the construction of low-rise residential units; however, a change in designation is possible to a partially residential and partially commercial nature. 3. Sėlių Street Lot has an overall area of 12.53 ares. The designation of this lot involves use within this location of residential and dormitory facilities. All city utility lines have been installed. Its great advantage is its location within the city center which has a beautifully developed infrastructure. Kindergartens, schools, stores, a polyclinic

3.4. Models for determining the most rational real estate purchase alternative and negotiations based on real calculations A real estate agent offers a package of services: showing a property, advising sellers on how to make the house more marketable, assessing current market conditions, providing information about property values and neighborhoods, matching buyers with sellers, negotiating the sale 8

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

Fig. 7. Correlation (recognition level) linking heart rate of potential real estate buyers with their interest in video clip content.

application of the Expert Estimate Method. The table used in analyzing these lots (criteria system, criteria weight and measuring units and significances) appears on the website: http://iti.vgtu.lt/VGTU_ Lomonosov/simpletable.aspx?sistemid=805. Descriptions of the alternatives under consideration and the criteria appear in greater detail on the above website. The selection of the alternatives or the criteria indicators in this website shows their detailed descriptions. A multiple criteria decision analysis of alternatives is performed based on the typical Capital PRO multiple criteria decision analysis matrix. The results show that the J. Jasinskio St. lot is assessed as the best, according to the compiled criteria system, with a degree of utility of 100%. The assessment of Alternative 2, the lot on Kęstučio St., puts it in last place, with a degree of utility of 70.33%.

and public transport stops are nearby. The accessibility of this lot is obviously better than the other alternatives. 4. Kauno Street Lot has an area of 11.20 ares. The nature of its use involves a residential location for the construction of low-rise residential units. Its great advantage is that, according to the master plan for the territory of the Vilnius City Municipality, this lot falls within the territory of district centers and other high-intensity construction sites. Thus, construction of multi-unit residential or commercial facilities is also possible on this site. There is an advantageous opportunity to expand development by acquiring neighboring lots. The infrastructure is fully developed—there are nearby kindergartens, schools, stores, a polyclinic and public transport stops. 5. J. Jasinskio Street Lot has an area of 5.27 ares, and is a location designated for public use. There is an approved, detailed plan. The lot's building intensity is 2.5 and its height is three stories. A deserted 380 m2 building stands on this site, for which a reconstruction permit has been authorized.

4.2. Establishing the emotional state of a potential buyer with the Microsoft Emotion API and QA5 SDK subsystems Worldwide experience shows (see Section 1) that a good portion of the decision making by a potential buyer depends on his/her emotional state while analyzing alternatives. For example, property buyers often reconsider their highest-priority selections numerous times. In this study, consideration of a potential buyer's emotional state while analyzing alternatives involved application of the Microsoft Emotion API and QA5 SDK subsystems (using eight criteria based on the analysis of face and voice emotion). The application of the Microsoft Emotion API subsystem involved showing a buyer the sequence of five different alternatives under assessment, displaying them every 12 seconds on average. Seven emotions of a potential buyer were assessed: happiness, neutrality, sadness, anger, surprise, fear and disgust. Table 3 shows that

The lots under discussion are as suitable for constructing housing as for commercial office buildings. One of the potential buyers is the company's owner, who needs a lot to construct new commercial facilities. He plans to lease part of these commercial facilities to other companies. The compiled system of criteria, which comprehensively describes the lots under consideration, is formed of two parts: the typical Capital PRO criteria system for assessing lots (see Table 2) and the criteria system describing the emotional state of a potential buyer during the time in which alternatives are analyzed (see Table 3). The weighted significance of the criteria were established using an

9

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

Table 2 Table for analyzing lots under consideration compiled by the typical Capital PRO multiple criteria analysis method and results of the multiple criteria assessment. Quantitative and qualitative information pertinent to alternatives a

Criteria describing the alternatives

Price Preliminary price per m2 for selling a possible building Lot area Visibility from the street Accessibility Infrastructure City center accessibility by automobile City center accessibility by public transport Building density Building intensity Height Parking lots/ability to rent from the municipality Utility lines Complexity of construction jobs Restrictions on digging jobs Demolition work on existing buildings Adjacent lots Detailed plan Building permit Attractiveness of the district Five-year perspectives Competitive environment Needed environmental investments (roads, paths, recreational zone) Sums of weighted, normalized, maximizing alternative indices (project Sums of weighted, normalized, maximizing alternative indices (project Significance of the alternative Priority of the alternative Utility degree of the alternative (%) a

Measuring units

− Euro/are + Euro/m2 + Ares + Points + Points + Points + Points + Points + % + Ratio + m + Points + Points − Points − Points − Points + Points + Points + Points + Points + Points − Points − Points “pluses”) “minuses”)

Weight

0.12 0.25 0.03 0.03 0.04 0.04 0.03 0.02 0.03 0.03 0.04 0.06 0.03 0.02 0.01 0.01 0.02 0.05 0.05 0.04 0.02 0.02 0.01

Lot alternatives under comparison Saltoniškių St.

Kęstučio St.

Sėlių St.

J. Jasinskio St.

Kauno St.

67,020 2100 10.37 4 4 4 5 4 50 3 22 0 3 3 1 0 3 0 0 4 4 4 1 0.1425 0.0401 0.1768 4 70.5%

48,413 2200 12.29 2 3 3 4 3 60 0.2 12 0 3 2 1 2 2 1 0 3 3 3 1 0.1338 0.0323 0.1764 5 70.33%

43,895 2200 12.53 4 4 4 4 4 50 2.4 18 1 3 3 1 2 2 1 0 3 3 3 1 0.1719 0.0318 0.2151 2 85.79%

98,861 2300 5.27 4 4 4 5 5 80 2.5 12 1 3 3 2 2 2 1 1 4 4 3 1 0.2261 0.0557 0.2508 1 100%

35,833 1900 11.2 3 3 3 4 4 80 1.6 16 1 3 3 2 2 1 0 0 2 3 2 2 0.136 0.0303 0.1814 3 72.33%

The sign + (−) indicates that a greater (lesser) criterion value corresponds to a greater (lesser) significance for stakeholders.

sad emotion, which comprises 85.32% of all emotions displayed for this site. There was also an analogous situation with Alternative 5. The sight of this lot caused a negative reaction, primarily since it was associated with the area of the train station; this was not very acceptable to potential buyers due to the negative outlook the public holds about this area. The faces of potential buyers displayed a sad emotion 10.63% of the time when reviewing this alternative. Alternative 4 took first place in terms of priority from the analysis of face emotions (see Table 3).

potential buyers almost never felt anger, surprise, fear or disgust. Alternatives under analysis (both from positive and negative aspects) shows that the emotions that affect further decision making are happiness, neutrality and sadness. Table 3 presents data on the emotional state of a potential buyer while reviewing the alternatives. For example, old buildings are present in Alternative 2; an old building on the lot creates negative emotions. Consequently, the face of the potential buyer predominantly displays a Table 3 Emotional state of a potential buyer while analyzing alternatives. Quantitative and qualitative information pertinent to alternatives Criteria describing the alternatives

*

Measuring units

Weight

Lot alternatives under comparison 1 Saltoniškių St.

1. Neutral 2. Happy 3. Sad 4. Angry 5. Surprised 6. Scared 7. Disgusted Rounded sum of all emotions equals 100% Alternative priorities assessed as per face emotional states More positive or more negative emotional states of a potential buyer while reading alternatives (see Table 4) 8. Alternative priorities assessed as per voice emotional states (see Table 4) Alternative priorities under analysis as per the typical, Capital PRO multiple criteria analysis method (see Table 2)

Microsoft emotion API affective database + % 0.03 54.27 + % 0.03 0.02 – % 0.03 43.46 – % 0.03 1.31 + % 0.03 0.01 – % 0.03 0.01 – % 0.03 0.22 % 100 4 +

QA5 SDK emotions database Points 0.03 −3 4

Table for analyzing lots under consideration 4

10

2 Kęstučio St.

3 Sėlių St.

4 J.Jasinskio St.

5 Kauno St.

13.27 0.01 85.32 0.72 0 0 0.08 100 5

54.98 36.87 6.74 0.04 0.02 0 0.04 100 2

2.21 97.63 0.11 0 0 0 0 100 1

83.91 4.82 10.63 0.03 0.02 0 0.06 100 3

−3

+3

+1

−1

4

1

2

3

5

2

1

3

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

Table 4 Recalculated voice analysis results a) digital and b) graphic forms.

a)

SAF

Atmosphere

Sum of positive emotions

Sum of negative emotions

Difference

Alternative Priority

↓↑

Extreme Emotion

↑

Imagination

↓

Intensive Thinking

↑

Embarrassment

↑

Brain Power

↓

Hesitation

↑

EmoCog Ratio

↓

Concentrated

↑

Excited

↑

Uncertain

↓

Stressed

↓↑

Angry

↓

Upset

↓

Content

↓

Energy

↑

Alternative

↓↑

1 2 3 4 5

↓1 ↓1 ↑2 ↑2 ↑2

0 0 0 0 0

1 1 1 1 1

0 0 0 0 0

3 4 2 2 2

1↓ 1↓ 3↑ 1↓ 3↓

5 6 6 8 5

5 3 3 5 2

1 1 1 1 1

6 6 6 5 7

2 2 2 2 2

1 1 1 2 1

1 3 1 1 1

1 1 1 1 1

1 1 1 2 1

1 1 2 1 1

1↓ 1↓ 1↑ 1↑ 1↓

7 7 9 8 7

8 8 6 7 8

-3 -3 3 1 -1

4 4 1 2 3

Negative emotions

Positive emotions

b)

third place. The analysis of facial and vocal emotions showed that potential buyers emotionally preferred Alternatives 3 (due to its better driving accessibility) and 4 (near to the center's Gedimino Prospect, one of the most attractive and sought-after streets in Vilnius). The faces of potential buyers indicated a happy emotion 36.87% of the time while reviewing Alternative 3 and 97.63% of the time while reviewing Alternative 4. The sum total of all the emotions is 100%. The sufficiently strong emotion of happiness indicated an optimistic outlook on Alternatives 3 and 4 by potential buyers. The analysis of a potential buyer's vocal emotion shows that Alternative 3 took first place as a priority, whereas Alternative 4 took second place (see Table 3). Table 3 presents an analysis of the lots under consideration using Subsystem QA5 SDK and the results of the typical Capital PRO multiple criteria decision analysis matrix. Fig. 8 presents a graphic representation of facial emotions. It can be noticed that, while looking at the first alternative, a potential buyer felt both neutral and sad, the levels of which fluctuated between 40%–60%. As a potential buyer watched Alternative 2, the emotional state of sadness reached a limit of 84%–93%. It can be asserted that the buyer did not like this particular land lot. Meanwhile, neutral and happy states dominated when analyzing the results of facial emotions for viewings of Alternative 3, while sadness fell to between 0% and 8.05%. Alternative 4 created the strongest emotion of happiness; the buyer was 97.63% happy while viewing it. The best of all the alternatives is Lot 4, as assessed based on the results of the emotion analysis. A neutral state fluctuated at 81%–93% while analyzing Alternative 5, whereas sadness comprised 6%–14%. Upon completing the analysis of facial emotions, an average for each of the seven emotions was calculated for each viewed alternative (happiness, neutrality, sadness, anger, surprise, fear and disgust). These appear in Table 3.

A buyer reads the descriptions of the alternatives under assessment sequentially, from the first to the fifth. The Voice Analysis Subsystem analyzes the vocal emotions of a buyer using 17 parameters (QA5 SDK…, 2009): Energy, Content, Upset, Angry, Stressed, Uncertain, Excited, Concentrated, EmoCog Ratio, Hesitation, Brain Power, Embarrassment, Intensive Thinking, Imagination Activity, Extreme Emotion, SAF and Atmosphere (see Table 4). The indicators in Table 4 show the direction of an emotion (↑ - the greater the value, the better the positive feeling; ↓ - the greater the value, the stronger the negative feeling). The numbers of positive and negative emotions were totaled separately, and the difference between them was noted. For example, Alternative 1 consisted of six positive emotions (Concentrated, EmoCog Ratio, Brain Power, Intensive Thinking, Imagination and SAF) and nine negative emotions (Energy, Upset, Stressed, Uncertain, Excited, Hesitation, Embarrassment, Extreme Emotion and Atmosphere). The difference between them was therefore three. Analogous calculations were performed with all the other alternatives (see Table 4). It can be seen from Table 4 (showing graphic and digital expressions) above that the buyer felt very similarly about Alternatives 1 and 2. The difference between the positive and the negative emotions regarding these alternatives was − 3. Therefore, these alternatives take fourth place. The results of the emotions in the voice analysis of Alternative 3 show predominately positive emotions; the difference between the positive and negative emotions was three. Therefore, the assessment of this alternative puts it in first place. The feelings of the buyer were somewhat more negative while reading Alternative 4 than they were while reading Alternative 3. These results are the basis for assessing this alternative in second place. The results of the emotions in the voice analysis of Alternative 5 show that the feelings of the potential buyer were somewhat more negative while reading Alternative 4 than they were while reading Alternatives 1 or 2. The difference between the positive and the negative emotions regarding Alternative 5 was −1. These results form the basis for assessing this alternative as being in 11

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

Fig. 8. Graphic representation of facial emotions.

Alternative 1 – 4th place

Alternative 2 – 5th place

Alternative 3 – 2nd place

Alternative 4 – 1st place

Alternative 5 – 3rd place

satisfied. Based on an analysis of the most advanced research worldwide, it can be shown that there are three advanced research contributions which are unique to the proposed method and NEAR:

4.3. Neuro decision matrix of data describing lot alternatives and results of the multiple criteria assessment of land lots A compilation of data is based on the methodology of typical Capital PRO multiple criteria decision analysis (see Table 2) which establishes the emotional state of a potential buyer while analyzing the alternatives (see Table 3). This compilation is a neuro decision matrix (see Table 5, http://iti.vgtu.lt/VGTU_Lomonosov/simpletable.aspx?sistemid=816) which comprehensively describes the lot alternatives. A multiple criteria decision analysis of alternatives is performed based on the compiled neuro decision matrix with results that appear in Table 5. The results show that the J. Jasinskio St. lot was assessed as the best according to the compiled criteria system. The degree of this lot's utility is 100%. The assessment of Alternative 2, the lot on Sėlių St., puts it in last place, with a degree of utility of 60.17%.

• Firstly, the method involves the compilation of data for a neuro •

5. Conclusion and future work Nobel Laureate Simon (1967, 1983) initiated a new era in decision theory when he introduced constrained rationality. Simon (1967) proposed a theory of the relationship of motivation and emotional behavior to information processing behavior. This theory explains how a basic serial information processor endowed with multiple needs behaves adaptively and survives in an environment that presents unpredictable threats and opportunities. For example, an interruption mechanism, that is, an emotion, allows the processor to respond to urgent needs in real time (Simon, 1967). Later, research in the areas of behavioral operational research and emotions in decision making conducted by other scientists (Chick et al., 2017; Garg and Lerner, 2013; George and Dane, 2016; Hämäläinen et al., 2013; Koshkaki and Solhi, 2016; Leon et al., 2017; Mikels et al., 2015; Petropoulos et al., 2016; Rick et al., 2014; So et al., 2015; Treur and Umair, 2015; Wolf et al., 2015; Yoo et al., 2016) as well as by the authors of this article has also shown that decision making depends on the emotional state of a person while analyzing alternatives. Globally, there are developments of multimodal (Lee and Norman, 2016; Esposito et al., 2015; Poria et al., 2017; Poria et al., 2016; Dai et al., 2015; Gauba et al., 2017; Chen et al., 2016; Mehmood et al., 2016; Ringeval et al., 2015) video biometric and affective computing systems, which attempt to analyze a user's emotions. The authors of this article applied this context in carrying out research relevant to the development of the proposed method and the NEuroAdvertising Property Video Recommendation System (NEAR). The results of this research show that the initial expectations were fully

•

decision matrix based on housing attributes, valence, arousal, emotional state and physiological parameters of a potential real estate buyer. Secondly, a multiple criteria neuroanalysis is carried out. This is followed by the selection of the most personalized and effective video clip ad from numerous alternative variants by considering the aforementioned neuro decision matrix and the features of a potential real estate buyer (e.g. age, gender, education, income and marital status). The basis for establishing the priorities, degrees of utility and market values of the variants under comparison consists of this neuro decision matrix and the application of methods for conducting multiple criteria analyses of projects, all developed by the authors of this article (Kaklauskas, 1999, 2016). Following this, electronic negotiations are made possible. The third innovation arises from the opportunities provided by the NEAR for presenting the most effective video clip ads for specific potential real estate buyers for as long as possible, according to Multiple Resource Theory.

These are the main integrated innovations of the proposed NEuroAdvertising Property Video Recommendation System (NEAR). As can be seen from the worldwide research reviewed in the Introduction, existing work in various areas has achieved significant results. However, nowhere in the world has a product as integrated as the NEAR previously been developed. 5.1. Practical implications This system can provide an aid in practical settings for streamlining the advertising process, raising the quality of video ad clips and making them more interesting and effective. In practice, the application of the NEAR can generate the necessary conditions for rationalizing the advertising process, improving the quality of advertising and increasing interest in the advertising. In this way, advertising can be personalized and made more suitable to the user's needs. An advertisement can more appropriately react to the constantly changing moods of its viewers; i.e. the system automatically selects and shows the most effective 12

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

Table 5 Neuro decision matrix of preliminary data comprehensively describing lot alternatives and results of the multiple criteria assessment of land lots. Quantitative and qualitative information pertinent to alternatives Criteria describing the alternatives

Neutral Happy Sad Angry Surprised Scared Disgusted Assessment of emotions by voice Price Preliminary price per m2 for selling a possible building Lot area Visibility from the street Accessibility Infrastructure City center accessibility by auto City center accessibility by public transport Building density Building intensity Height Parking lots/ability to rent from the municipality Utility lines Complexity of construction jobs Restrictions on digging jobs Demolition work on existing buildings Adjacent lots Detailed plan Building permit Attractiveness of the district Five-year perspectives Competitive environment Needed environmental investments (roads, paths, recreational zone) Sums of weighted, normalized, maximizing alternative indices (project Sums of weighted, normalized, maximizing alternative indices (project Significance of the alternative Priority of the alternative Utility degree of the alternative (%) a

a

Measuring units

+ % + % − % − % + % − % − % − Priority − Euro/are + Euro/m2 + Ares + Points + Points + Points + Points + Points + % + Ratio + m + Points + Points − Points − Points − Points + Points + Points + Points + Points + Points − Points − Points “pluses”) “minuses”)

Weight

0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.12 0.25 0.03 0.03 0.04 0.04 0.03 0.02 0.03 0.03 0.04 0.06 0.03 0.02 0.01 0.01 0.02 0.05 0.05 0.04 0.02 0.02 0.01

Compared alternatives Saltoniškių St.

Kęstučio St.

Sėlių St.

J. Jasinskio St.

Kauno St.

54.27 0.02 43.46 1.31 0.01 0.01 0.22 4 67,020 2100 10.37 4 4 4 5 4 50 3 22 0 3 3 1 0 3 0 0 4 4 4 1 0.1563 0.1228 0.1882 4 60.18%

13.27 0.01 85.32 0.72 0 0 0.08 4 48,413 2200 12.29 2 3 3 4 3 60 0.2 12 0 3 2 1 2 2 1 0 3 3 3 1 0.1357 0.0747 0.1882 5 60.17%

54.98 36.87 6.74 0.04 0.02 0 0.04 1 43,895 2200 12.53 4 4 4 4 4 50 2.4 18 1 3 3 1 2 2 1 0 3 3 3 1 0.1997 0.0389 0.3005 2 96.09%

2.21 97.63 0.11 0 0 0 0 2 98,861 2300 5.27 4 4 4 5 5 80 2.5 12 1 3 3 2 2 2 1 1 4 4 3 1 0.2474 0.06 0.3128 1 100%

83.91 4.82 10.63 0.03 0.02 0 0.06 3 35,833 1900 11.2 3 3 3 4 4 80 1.6 16 1 3 3 2 2 1 0 0 2 3 2 2 0.1611 0.0438 0.2507 3 80.14%

The sign + (−) indicates that a greater (lesser) criterion value corresponds to a greater (lesser) significance for stakeholders.

Development of NEAR2 (Analysis and Assessment of Electronic Advertising Influences [ad contents under development] System) should be the third direction of this future research. NEAR2 would permit more to be learned about the effectiveness of an advertisement during each stage of its development. It would assist in establishing the strengths and weaknesses of an advertisement and the means for improving it, so that the ad would become the maximally attractive variant for a certain viewer. This system would assist in establishing the part of the video with the greatest influence on short-term and longterm memory and the items in the clip which cause the strongest impressions and feelings. For example, application of this system would allow determination of the number of times the repetition of certain news in a specific area of the video is necessary for a successful advertising campaign. This sort of information makes it possible to interchange the elements of a video clip to make the transmitted news more visible and memorable. Development of Integrated Cultural Heritage Analytics (CHAT) based on NEAR is expected to constitute the fourth direction for future research. CHAT will consist of video neuroanalytics and opinion analytics. Video neuroanalytics will be used to analyze the beauty of cultural heritage objects (buildings, monuments, open spaces, streets, cultural spaces) and cultural events, taking into account a viewer's valence, arousal and emotional state, and to measure, analyze and rate the cultural heritage object according to the viewer's valence, arousal and emotional state (pleasure, displeasure etc.) and generate recommendations for effectively managing these spaces and cultural

advertisement in accordance with a viewer's emotions. Three directions can be identified for the practical implications of the NEAR. The initial expectation is the widespread installation of NEAR within Lithuania's real estate and construction organizations. Additionally, use of NEAR is expected in other industries (e.g. construction, banking, fashion, food, healthcare and service industries). Furthermore, development of the Integrated Cultural Heritage Analytics (CHAT) based on NEAR will be used to implement the Horizon 2020 project, “Regeneration and Optimization of Cultural Heritage in Creative and Knowledge Cities”. CHAT will consist of video neuroanalytics and opinion analytics. 5.2. Future research directions The NEAR system as developed here is not ideal; it does have certain limitations. Future research on the NEAR is foreseen as taking five directions. Firstly, the existing databases and the model-base under adaptation will be extended; this will enable the use of NEAR within other industries (construction, banking, fashion, food, healthcare and service industries). In this case, data from these industries will supplement the existing Property Video Clip Ads Database, Personalized Questionnaire and Answers Database and Real Estate Database. The names of these Databases will change in parallel. The second direction for future research would involve an expansion of the possibilities of NEAR by applying the vector and Positive Activation– Negative Activation (PANA) two-dimensional models. The development of NEAR involved application of Russell's (1980) circumplex model of emotion. 13

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

Belch, G.E., Belch, M.A., 2012. Advertising and Promotion: An Integrated Marketing Communications Perspective. McGraw-Hill Irwin, New York. Cash, P., Holm-Hansen, C., Olsen, S.B., Christensen, M.L., Trinh, Y.M.T., 2017. Uniting individual and collective concerns through design: priming across the senses. Des. Stud. 49, 32–65. Challcharoenwattana, A., Pharino, C., 2016. Multiple-criteria decision analysis to promote recycling activities at different stages of urbanization. J. Clean. Prod. 137, 1118–1128. Chen, X., An, L., Yand, S., 2016. Zapping prediction for online advertisement based on cumulative smile sparse representation. Neurocomputing 175 (A), 667–673. Chick, C.F., Pardo, S.T., Reyna, V.F., Goldman, D.A., 2017. Decision making (individuals). In: Ref. Mod. in Neurosci. Biobehav. Psychol, http://dx.doi.org/10.1016/B978-0-12809324-5.06393-8. Dahl, S., Gordon-Wilson, S.M., 2013. Advertising assertiveness and effectiveness: the role of product involvement. In: Proceedings of the Academy of Marketing, Marketing Relevance. Glamorgan, Wales (UK), Academy of Marketing, pp. 1–6. Dai, W., Han, D., Dai, Y., Xu, D., 2015. Emotion recognition and affective computing on vocal social media. Inf. Manag. 52 (7), 777–788. Deaconu, A., Lazar, D., Buiga, A., Fatacean, G., 2016. Marginal prices of improvements made to blocks of flats: empirical evidence from Romania. Int. J. Strateg. Prop. Manag. 20 (2), 156–171. Esposito, A., Esposito, A.M., Vogel, C., 2015. Needs and challenges in human computer interaction for processing social emotional information. Pattern Recogn. Lett. 66, 41–51. Ferreira, F.A.F., Jalali, M.S., 2015. Identifying key determinants of housing sales and time-on-the-market (TOM) using fuzzy cognitive mapping. Int. J. Strateg. Prop. Manag. 19 (3), 235–244. Ferreira, F.A.F., Marques, C.S.E., Bento, P., Ferreira, J.J.M., Jalali, M.S., 2015. Operationalizing and measuring individual entrepreneurial orientation using cognitive mapping and MCDA techniques. J. Bus. Res. 68 (12), 2691–2702. Ferreira, F.A.F., Jalali, M.S., Ferreira, J.J.M., 2016a. Experience-focused thinking and cognitive mapping in ethical banking practices: from practical intuition to theory. J. Bus. Res. 69 (11), 4953–4958. Ferreira, F.A.F., Jalali, M.S., Ferreira, J.J.M., 2016b. Integrating qualitative comparative analysis (QCA) and fuzzy cognitive maps (FCM) to enhance the selection of independent variables. J. Bus. Res. 69 (4), 1471–1478. Garg, N., Lerner, J.S., 2013. Sadness and consumption. J. Consum. Psychol. 23 (1), 106–113. Gauba, H., Kumar, P., Roy, P.P., Singh, P., Dogra, D.P., Raman, B., 2017. Prediction of advertisement preference by fusing EEG response and sentiment analysis. Neural Netw. http://dx.doi.org/10.1016/j.neunet.2017.01.013. George, J.M., Dane, E., 2016. Affect, emotion, and decision making. Organ. Behav. Hum. Decis. Process. 136, 47–55. Gohary, A., Hanzaee, K.H., 2014. Personality traits as predictors of shopping motivations and behaviors: a canonical correlation analysis. Arab Econ. Bus. J. 9 (2), 166–174. Gunter, B., Baluch, B., Duffy, L.J., Furnham, A., 2002. Children's memory for television advertising: effects of programme–advertisement congruency. Appl. Cogn. Psychol. 16 (2), 171–190. Hämäläinen, R.P., Luoma, J., Saarinen, E., 2013. On the importance of behavioral operational research: the case of understanding and communicating about dynamic systems. Eur. J. Oper. Res. 228 (3), 623–634. Hawkins, D.I., Mothersbaugh, D.L., Mookerjee, A., 2011. Consumer Behavior: Building Marketing Strategy, 11th ed. Tata McGraw-Hill, New York. Hilton Head Real Estate Partners, 2016. Where is housing headed for the rest of 2016? Available: http://hiltonheadrealestatepartners.com/hilton-head-real-estate-partners/ blog/ (web archive link, 18 December 2016) (accessed on 18 December 2016). Jayantha, W.M., Lau, J.M., 2016. Buyers' property asset purchase decisions: an empirical study on the high-end residential property market in Hong Kong. Int. J. Strateg. Prop. Manag. 20 (1), 1–16. Kahneman, D., 2011. Thinking, Fast and Slow. Macmillan (499 pp.). Kaklauskas, A., 1999. Multiple criteria decision support of building life cycle. In: Research Report Presented for Habilitation (DrSc): Technological Sciences, Civil Engineering (02T), Vilnius Gediminas Technical University. Technika, Vilnius. Kaklauskas, A., 2016. Degree of project utility and investment value assessments. Int. J. Comput. Commun. Control 11 (5), 667–684. Kaklauskas, A., Rute, J., Zavadskas, E.K., Daniūnas, A., Pruskus, V., Bivainis, J., Gudauskas, R., Plakys, V., 2012. Passive house model for quantitative and qualitative analyses and its intelligent system. Energ. Buildings 50, 7–18. Kaklauskas, A., Kuzminske, A., Ubarte, I., 2015a. Biuro darbuotojų multimodalinių biometrinių parametrų, priklausomybės nuo streso mokslinis taikomasis tyrimas. In: Priemonė Inočekiai LT. VP2–1.3-ŪM-05-K-03-858, Applicant: UAB MD Projects, Executor: Vilnius Gediminas Technical University. Kaklauskas, A., Kuzminske, A., Ubarte, I., 2015b. Biuro darbuotojų darbo našumo priklausomybės nuo multimodalinių biometrinių temperatūros duomenų tyrimas. In: Priemonė Inočekiai LT. VP2–1.3-ŪM-05-K-03-851, Applicant: UAB Baltic Business Solutions, Executor: Vilnius Gediminas Technical University. Kanapeckiene, L., Kaklauskas, A., Zavadskas, E.K., Raslanas, S., 2011. Method and system for multi-attribute market value assessment in analysis of construction and retrofit projects. Expert Syst. Appl. 38 (11), 14196–14207. Koshkaki, E.R., Solhi, S., 2016. The facilitating role of negative emotion in decision making process: a hierarchy of effects model approach. J. High Technol. Managem. Res. 27 (2), 119–128. Kvietkutė, R., 2013. Motivation for Consumers to Buy Luxury Brands. (Master thesis) Vytautas Magnus University, Kaunas, Lithuania (in Lithuanian). Lee, W., Norman, M.D., 2016. Affective computing as complex systems science. Procedia Comput. Sci. 95, 18–23.

events, attracting residents, businesses, students, tourists, cultural operators and events and others. Different stakeholders can receive advice on ways to improve the sustainability, effective regeneration and adaptive reuse of a particular cultural heritage asset. Opinion analytics will be used to enable the automatic detection of opinions expressed in articles, reviews, surveys, comments, opinions, notices, papers, research, studies, blogs, online forums, Facebook, Twitter and other social media channels, thereby allowing visualization of the opinions held by citizens on issues of urban cultural heritage. It will include the use of keyword dictionaries to classify documents (articles, reviews, surveys, comments, opinions, notices, papers, research, studies, online forums and/or other social media channels) as positive, neutral or negative. The application of opinion analytics will allow better understanding and monitoring of opinions, thoughts, sentiments, attitudes, emotions and preferences of urban citizens and allow city officials to make decisions that are more advanced. Opinion analytics will summarize the opinions of urban citizens from a range of social media channels and identify trends regarding the cultural heritage, in order to foster more creative and knowledgeable cities, and will help cities to better measure and understand the opinions of citizens on key urban cultural heritage issues. The innovativeness of CHAT will be important in automatically determining the “temperature” of the sustainability of cultural heritage, compiling numerous alternative recommendations applicable to a specific user, performing a multiple criteria analysis of these recommendations and selecting the ten most appropriate options for that user. Likewise, NEAR only analyzes certain socioeconomic status demographics of buyers (age, gender, education, income, occupation, family structure and social class) from all possible individual differences of a potential property buyer; personality, personal values and lifestyle, motivation, knowledge, intention, attitudes, beliefs and feelings are left aside for now, possibly for future analysis. A future fifth direction for this research is to add a feature allowing users to choose for themselves which environmental influences, individual differences and housing attributes should be considered in the sorting of alternative video ads of properties for sale. The purpose of the sixth direction of future work is to integrate NEAR with the analysis of inaudible sound, using an E4 wristband or similar wristbands intended for analyzing emotions as well as Google Glass, Picard Smart Watch or similar glasses intended for analyzing emotions and other MCDA techniques (Ferreira et al., 2015, 2016a, 2016b). This would permit a more effectively personalized, affective computing process for a viewer which takes into account the available past and real-time data. Moreover, an improved integration of affective data gained from viewers interacting with NEAR is predicted in the future from the synchronization of different modalities. Acknowledgements The authors thank the referees for their valuable comments and suggestions, which helped to improve this paper considerably. References Alkan, L., 2015. Housing market differentiation: the cases of Yenimahalle and Çankaya in Ankara. Int. J. Strateg. Prop. Manag. 19 (1), 13–26. Alkay, E., 2015. Housing choice structure: examples of two different-size cities from Turkey. Int. J. Strateg. Prop. Manag. 19 (2), 123–136. Bakalash, T., Riemer, H., 2013. Exploring ad-elicited emotional arousal and memory for the ad using fMRI. J. Advert. 42 (4), 275–291. Barrett, L.F., Mesquita, B., Gendron, M., 2011. Context in emotion perception. Curr. Dir. Psychol. Sci. 20 (5), 286–290. Bastiaansen, M., Straatman, S., Driessen, E., Mitas, O., Stekelenburg, J., Wang, L., 2016. My destination in your brain: a novel neuromarketing approach for evaluating the effectiveness of destination marketing. J. Destin. Mark. Manag. http://dx.doi.org/10. 1016/j.jdmm.2016.09.003. Belanche, D., Flavián, C., Pérez-Rueda, A., 2017. Understanding interactive online advertising: congruence and product involvement in highly and lowly arousing, skippable video ads. J. Interact. Mark. 37, 75–88.

14

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al. Leon, R.-D., Rodríguez-Rodríguez, R., Gómez-Gasquet, P., Mula, J., 2017. Social network analysis: a tool for evaluating and predicting future knowledge flows from an insurance organization. Technol. Forecast. Soc. Chang. 114, 103–118. Martins, V.C.S., Filipe, M.N.M., Ferreira, F.A.F., Jalali, M.S., António, N.J.S., 2015. For sale… but for how long? A methodological proposal for estimating time-on-themarket. Int. J. Strateg. Prop. Manag. 19 (4), 309–324. Maslow, A.H., 1943. A theory of human motivation. Psychol. Rev. 50 (4), 370–396. McLaughlin, R., 2015. Dreaming big: Americans still yearning for larger homes. Available: http://www.trulia.com/blog/trends/americans-larger-homes/ (web archive link, 12 October 2016) (accessed on 12 October 2016). Mehmood, I., Sajjad, M., Rho, S., Baik, S.W., 2016. Divide-and-conquer based summarization framework for extracting affective video content. Neurocomputing 174 (A), 393–403. Mikels, J.A., Shuster, M.M., Thai, S.T., 2015. Aging, emotion, and decision making. In: Hess, T.M., Strough, J., Löckenhoff, C. (Eds.), Aging and Decision Making: Empirical and Applied Perspectives. CA: Elsevier Academic Press, San Diego, pp. 169–188. Morgan, R., Media, D., 2016. Difference in marketing strategy towards men & women. Available: http://smallbusiness.chron.com/difference-marketing-strategy-towardsmen-women-15438.html (web archive link, 01 December 2016) (accessed on 1 December 2016). NAR, 2015. Home buyer and seller generational trends. Available: http://www.realtor. org/sites/default/files/reports/2015/2015-home-buyer-and-seller-generationaltrends-2015-03-11.pdf (web archive link, 01 December 2016) (accessed on 1 December 2016). National Association of Realtors, 2016. Home buyer and seller generational trends. Available. http://www.realtor.com/news/real-estate-news/generational-homebuying-trends/ (web archive link, 08 December 2016) (accessed on 8 December 2016). Pant, P., 2012. Should you invest in this rental property? Available: http:// affordanything.com/2012/01/25/income-property/ (web archive link, 08 December 2016) (accessed on 8 December 2016). Petropoulos, F., Fildes, R., Goodwin, P., 2016. Do 'big losses' in judgmental adjustments to statistical forecasts affect experts' behaviour? Eur. J. Oper. Res. 249 (3), 842–852. Poria, S., Cambria, E., Howard, N., Huang, G.-B., Hussain, A., 2016. Fusing audio, visual and textual clues for sentiment analysis from multimodal content. Neurocomputing 174 (A), 50–59. Poria, S., Cambria, E., Bajpai, R., Hussain, A., 2017. A review of affective computing: from unimodal analysis to multimodal fusion. Inform. Fusion 37, 98–125. QA5 SDK v. 5.5. Product Description & User Guide Ltd., Nemesysco, 2009. Available: http://www.nemesysco.com/partners2/Manuals/QA5/QA5%20SDK%20Rel5. 5%20March09_V1.pdf (accessed on 22 December 2016). Qin, J., Liu, X., Pedrycz, W., 2017. An extended TODIM multi-criteria group decision making method for green supplier selection in interval type-2 fuzzy environment. Eur. J. Oper. Res. 258 (2), 626–638. Quester, P., Pettigrew, S., Hawkins, D.I., 2011. Consumer Behavior: Implications for Marketing Strategy, 6th ed. McGraw-Hill Australia Pty Limited, Sydney. Reeves, R.V., 2015. The dangerous sepation of the American upper middle class. Available: http://www.brookings.edu/blogs/social-mobility-memos/posts/2015/ 09/03-separation-upper-middle-class-reeves (web archive link, 20 October 2016) (accessed on 20 October 2016). Rick, S.I., Pereira, B., Burson, K.A., 2014. The benefits of retail therapy: making purchase decisions reduces residual sadness. J. Consum. Psychol. 24 (3), 373–380. Ringeval, F., Eyben, F., Kroupi, E., Yuce, A., Thiran, J.-P., Ebrahimi, T., Lalanne, D., Schuller, B., 2015. Prediction of asynchronous dimensional emotion ratings from audiovisual and physiological data. Pattern Recogn. Lett. 66, 22–30. Romero, D., Bezdeka, T., 2014. Understandind 2014 C.A.R.® Home Buyer Survey. Available: http://www.slideshare.net/century21award/car-home-buyer-surveyseminar (web archive link, 18 December 2016) (accessed on 18 December 2016). Russell, J., 1980. A circumplex model of affect. J. Pers. Soc. Psychol. 39, 1161–1178. Simon, H.A., 1967. Motivational and emotional controls of cognition. Psychol. Rev. 74 (1), 29–39. Simon, H.A., 1983. Reason in Human Affairs. Stanford University Press, Stanford, CA. Simon, H., 1997. Administrative Behavior, 4th ed. The Free Press, New York. So, J., Achar, C., Han, D., Agrawal, N., Duhachek, A., Maheswaran, D., 2015. The psychology of appraisal: specific emotions and decision-making. J. Consum. Psychol. 25 (3), 359–371. Trapasso, C., 2016. Why single women's homes appreciate less than single men's. Available: http://www.realtor.com/news/real-estate-news/single-womens-homesappreciate-less/ (web archive link, 18 December 2016) (accessed on 18 December 2016). Treur, J., Umair, M., 2015. Emotions as a vehicle for rationality: rational decision making models based on emotion-related valuing and Hebbian learning. In: Biologically Inspired Cognitive Architectures 14. pp. 40–56. Wang, F., Zhang, P., Shang, Y., Shi, Y., 2013. The application of multiple criteria linear programming in advertisement clicking events prediction. Procedia Comput. Sci. 18, 1720–1729. Wilson, P., Wilson, J., 2015. Married Vs. single home buyers. Available: http://www. greathomesinalbuquerque.com/Blog/Married-Vs-Single-Home-Buyers (web archive link, 18 December 2016) (accessed on 18 December 2016). Wolf, I., Schröder, T., Neumann, J., de Haan, G., 2015. Changing minds about electric cars: an empirically grounded agent-based modeling approach. Technol. Forecast. Soc. Chang. 94, 269–285. Yoo, C.W., Goo, J., Derrick Huang, C., Nam, K., Woo, M., 2016. Improving travel decision support satisfaction with smart tourism technologies: a framework of tourist elaboration likelihood and self-efficacy. Technol. Forecast. Soc. Chang. http://dx.doi. org/10.1016/j.techfore.2016.10.071.

PhD Dr. Sc. Arturas Kaklauskas is a Professor at Vilnius Gediminas Technical University, Lithuania. His additional leadership duties are Director of the Research Institute of Smart Building Technologies, Head of the Department of Construction Economics and Property Management, laureate of the Lithuanian Science Prize, member of the Lithuanian Academy of Sciences and Editor of Engineering Applications of Artificial Intelligence, an international journal. He has contributed to nine Framework programmes, two HORIZON2020 projects and participated in over 30 other projects in the EU, US, Africa and Asia. The Belarusian State Technological University (Minsk, Belarus) awarded him an Honorary Doctorate in 2014. His publications include nine books and 121 papers in Web of Science Journals. Fifteen PhD students successfully defended their theses under his supervision. The Web of Science H-Index of Prof. A. Kaklauskas is 24. Web of Science Journals have cited him 1877 times. His areas of interest include multiple criteria decision analysis, intelligent decision support systems, affective computing, big data analytics, intelligent tutoring systems, intelligent library, internet of things, life cycle analyses of built environments, energy, climate change, resilience management, healthy houses, sustainable built environments, etc.

PhD Dr. Sc. Edmundas Kazimieras Zavadskas works as Professor and Head of the Department of Construction Technology and Management at Vilnius Gediminas Technical University, Lithuania. His additional duties are Senior research fellow at the Research Institute of Internet and Intelligent Technologies. He earned his PhD in Building Structures in 1973 and Dr. Sc. in Building Technology and Management in 1987. He is also a member of Lithuanian and several foreign Academies of Sciences, holds a Doctore Honoris Causa from the universities at Poznan, Saint Petersburg and Kiev and an Honorary International Chair Professor at the National Taipei University of Technology. He holds memberships in international organizations including steering and programme committees at many international conferences and editorial boards of several research journals. He has authored and co-authored more than 400 papers and numerous monographs in Lithuanian, English, German and Russian. He is Editor in Chief of Technological and Economic Development of Economy and Journal of Civil Engineering and Management. Research interests are building technology and management, decisionmaking theory, automation in design and decision support systems. Dr. Audrius Banaitis is a Professor in the Department of Construction Economics and Property Management at Vilnius Gediminas Technical University, Lithuania. His research interests include project/risk management and project success, property management, sustainability and construction industry development, innovation management, strategic management, multiple criteria decision making: applications in construction and real estate. He is the Editor-in-Chief of the International Journal of Strategic Property Management. He took part in numerous national and international R & D and education projects. Dr. Ieva Meidute-Kavaliauskiene is an Associate Professor at Vilnius Gediminas Technical University in Lithuania and at BRU-IUL, University Institute of Lisbon in Portugal. Her PhD degree is in Logistics and Supply chain management from the Faculty of Business Management at Vilnius Gediminas Technical University in Lithuania. Her research interests include data analysis, management, theory, entrepreneurship, elearning, SPSS, public speaking, academic writing, business strategy, project management, management consulting, strategic planning, distance learning, courses and leadership. Amir Liberman is a worldwide leading researcher in the field of human voice analysis. His more recent discoveries and additional novel vocal parameters were published in his second voice analysis patent from 1999, identifying “Concentration”, “Anticipation” and “Arousal” (also known as the “Love Detector” patent). Being a self-educated researcher, Amir's methods of research are very unique and unfamiliar to the world of traditional phonetics, as most of his research is done in real-life settings and not in a mocked laboratory atmosphere. Amir formed Nemesysco, Ltd. in April of 2000, to manage all of his IP patents and development projects. Since then, he has invested all of his efforts and resources in perfecting and fine-tuning LVA technology and its applications for home-land security needs, fraud prevention solutions, call centers utilities and CRM appliances. Simona Dzitac holds a Ph.D. in Engineering and lectures at Energy Engineering Faculty, University of Oradea in Romania. The degrees she has earned are a B.Sc. in 2000, M.Sc. in Mathematics-Physics in 2001, B.Sc. in 2005, M.Sc. in 2007 and a PhD in Energy Engineering from University of Oradea in 2008 and B.Sc. in Economic Informatics from University of Craiova in Romania in 2007. Her current research interests include reliability and applied mathematics and computer science in engineering fields. She has published 10 books and over 70 scientific papers in journals and conferences proceedings. Ieva Ubarte is a Ph.D. student at the Department of Construction Technology and Management and a Junior Researcher at the Research Institute of Smart Building Technologies at Vilnius Gediminas Technical University, Lithuania. Her research interests include decision support systems, sustainable built environment, safe and healthy house. Arune Binkyte is a Ph.D. student at the Department of Construction Economics and Property Management and a Researcher at the Research Institute of Smart Building Technologies at Vilnius Gediminas Technical University, Lithuania. Her research interests include sustainable construction, building certification, green building, multi-building assessment, the latest construction and the built environment technology and big data analysis.

15

Technological Forecasting & Social Change xxx (xxxx) xxx–xxx

A. Kaklauskas et al.

biosignal subsystems applied to detect stress, labour productivity and different emotions by means of physiological and behavioural signals; development and application of Webbased intelligent systems.

Justas Cerkauskas is a Ph.D. student at the Department of Construction Economics and Property Management and a Researcher at the Research Institute of Smart Building Technologies at Vilnius Gediminas Technical University, Lithuania. His research interests include text analytics, text mining, machine learning and recommender systems.

Andrej Naumcik is a Ph.D. student at the Department of Construction Technology and Management at Vilnius Gediminas Technical University, Lithuania. He is a co-owner of Capital PRO, a real estate broker company. His research interests include real estate, negotiation, real estate transactions, business strategy, investment properties, real estate development and buyer representation.

Dr. Agne Kuzminske is working as a lecturer at Department of Construction Economics and Property Management and as a researcher at the Research Institute of Smart Building Technologies in Vilnius Gediminas Technical University. She is working in the research fields related to: application of biometric technologies (physiological, behavioural),

16