TrailTrade: A model for trail-aware commerce support

Computers in Industry 80 (2016) 43–53

Contents lists available at ScienceDirect

Computers in Industry journal homepage: www.elsevier.com/locate/compind

TrailTrade: A model for trail-aware commerce support Jorge Luis Victória Barbosaa,* , Cládio Martinsa , Laerte Kerber Francoa , Débora Nice Ferrari Barbosab a b

University of Vale do Rio dos Sinos (UNISINOS), 950, Unisinos Av., São Leopoldo, 93.022-000, RS, Brazil FEEVALE University, 2755, ERS-239, Novo Hamburgo, 93.352-000, RS, Brazil

A R T I C L E I N F O

Article history: Received 24 June 2015 Received in revised form 22 January 2016 Accepted 14 April 2016 Available online 2 May 2016 Keywords: Context awareness Context-aware commerce Trail awareness Trail-aware commerce

A B S T R A C T

Context-aware applications adapt their functionalities based on users’ profiles and their current contexts. Complementarily, a trail is the history of the contexts visited by a user. Trails enable applications to explore the users’ past behavior. In this sense, trail awareness is considered an evolution of the context awareness, because applications explore an additional and more complete information source. This article proposes a model for trail-aware commerce support, called TrailTrade. The model uses the dealers’ profiles, contexts and trails to find deal opportunities and promote business transactions. Dealers may be people or companies offering and/or looking for something. TrailTrade supports the trade in goods and the exchange or sale of experience or knowledge. The model was implemented including an indoor location system and a deal assistant on mobile devices. The prototype was installed in a real environment and used to simulate a city composed of houses, a metro and a mall. This environment allowed a practical application in a scenario. This test evaluated the TrailTrade’s functionalities, mainly its trail awareness support. The results were encouraging and show potential for applying TrailTrade in real situations, fostering negotiations through the past behavior of dealers. ã 2016 Elsevier B.V. All rights reserved.

1. Introduction The evolution of mobile devices and high-speed wireless networks has stimulated researches related to mobile computing [1–3]. In addition, the improvement and proliferation of location systems [4,5] have motivated the use of precise location to provide location-based services [6–9]. Beyond the location, additional information related to users and their contexts allow contextaware computing [10–12]. Based on contexts, adaptive applications [13] can modify their behavior to better support the users’ needs. Today, context-aware applications allow personalized services through users’ profiles and their current contexts [14,15]. However, recent researches on adaptive applications indicate the relevance of consider the users’ past actions in visited contexts [16,17] to make decisions. These actions can be applications used, contents accessed and business transactions conducted. This approach improves the adaptive services, because they use the history of users’ actions, rather than only their current contexts and profiles. The past contexts are referred to as contexts histories [18] or, in a

* Corresponding author. E-mail addresses: [email protected] (J.L.V. Barbosa), [email protected] (C. Martins), [email protected] (L.K. Franco), [email protected] (D.N.F. Barbosa). http://dx.doi.org/10.1016/j.compind.2016.04.006 0166-3615/ ã 2016 Elsevier B.V. All rights reserved.

short way, trails [19–21]. Dey et al. [10] briefly described the importance of using trails in the decision making process. Therefore, in this article trail awareness is considered an evolution of the context awareness. So, trail-aware applications are considered more capable than context-aware applications to make decisions and help the users in their purposes. Context-aware computing has been applied in several areas, such as, education [14,15,21], logistics [17], health [22], competence management [23] and commerce [24]. Context-aware commerce [25–27] applications manage contextualized deal opportunities and, in the same way that occurs in any contextaware application, the decisions consider only the dealers’ profiles and contexts. Applications dedicated to pervasive/ubiquitous commerce [24,28–38] also have the same restriction. Deal discovery guided by the dealers’ trails is still an open research area. So, this article proposes TrailTrade, a model to explore trails to find and promote deal opportunities. The article is organized as follows. Section 2 discusses works in two research areas, namely trails management and context-aware/ pervasive/ubiquitous commerce. In particular, the section highlights the scientific contribution of TrailTrade. The third section describes the model, with an emphasis on its architecture and strategy to support the trail awareness. Sections 4 and 5 address the aspects of

44

J.L.V. Barbosa et al. / Computers in Industry 80 (2016) 43–53

implementation and evaluation. Finally, Section 6 presents final remarks and directions for future research.

services for trails. Thus, UbiTrail was chosen for trails management in TrailTrade.

2. Related works

2.2. Context-aware and pervasive/ubiquitous commerce

This section revises background concepts and discusses related works. Two strategic topics are addressed, namely trails management and context-aware/pervasive/ubiquitous commerce. This literature review shows that TrailTrade is the first model of trail-aware commerce.

The use of mobile devices and pervasive/ubiquitous technologies [47–49] to support commerce is commonly classified as context-aware commerce [25–27] or pervasive/ubiquitous commerce [24,28–38]. Galanxhi-Janaqi and Nah [35] and Roussos et al. [36] show that u-commerce is based on the infrastructure and experience of e-commerce and m-commerce. U-commerce often involves the electronic identification of physical products and seamless provisioning of business and consumer services in ubiquitous computing infrastructures [47–49]. Gershman [37] indicates as prerequisites for the success of u-commerce: (1) always be connected with clients; (2) always be aware of clients’ contexts (where they are, what they are doing and what is available around them); and (3) always be proactive, identifying real-time opportunities to meet client needs. Some proposals focus on commerce in goods. iGrocer [30] is a smart assistant to aid clients in the purchase process in supermarkets. iGrocer uses consumers’ nutritional profiles to suggest the purchase of products or even warn about items that should be avoided. This is particularly practical for elderly people or those who need help to follow a specific diet. iGrocer also identifies the desired products through barcode, and gives the user the nutritional information. In order to suggest products for the consumer, the assistant also combines the client’s nutritional profile with a database of nutritional information. MyGrocer [33] is similar to iGrocer [30]. However, MyGrocer emphasizes the stock control of products in food-related businesses and households. MyGrocer uses sensors on the shelves and elsewhere to indicate the available quantity of a product, warning when something is missing or finishing. Lin et al. [28] propose a generic architecture called the Pervasive Activity Manager (PAM). PAM assists clients through complementary product information to help decision about buying an item. PAM uses services provided by the stores, malls, or even other customers. Other proposals focus on the services market. EPARK [29] assists clients and employees in the parking management process. Some of the features are display parking maps, book a parking space, and extend the booking time for a parking space. UTAS [38] uses a mobile device to assist tourists in situations where language is an obstacle. In addition, the system warns about delayed or canceled flights, obstructed roads, and restaurant or concert bookings. MUCS [24] uses the ubiquitous computing technologies to identify business opportunities. The system proposes a generic

2.1. Trails management Applications based on past contexts need to store them for further use. Sometimes, contexts history [18,39] is called trail [19– 21]. Bush [40] was the first researcher which employed the term trail to represent a history. Life logging [41–43] uses trails to enhance human memory through records of conversations, documents, location information, photos, audios, e-mails and videos, and many other types of personal and environmental data [44]. Trails management consists of techniques to store and use trails. Trails contain strategic information that can be used in advanced applications, such as recommendation systems [39], profile management [16] and prediction systems [45]. Trails can be used to personalize services and contents according to users’ previous choices [18,46]. Typically, the application is based on well-defined domain representation through an ontology [18,20] or an entityrelationship model [46]. The domain definition facilitates queries and reasoning to discover users’ preferences based on their past actions (trails). The literature review did not find any work addressing trail management or trail-awareness in the commerce domain. However, generic models for trails management which can be applied in any domain were found [18,20,41]. Trail-aware systems are based on trails management. Therefore, the generic models found support TrailTrade development. Life annotation [41] proposes the annotation of the places visited by users of mobile devices, thereby composing their trails. However, Life annotation [41] does not standardize trails information, which makes it difficult to query them. Hong et al. [18] proposes a mechanism to store trails based on an ontology. However, the work focuses only on techniques to take decisions. In addition, the proposal does not support the creation of services to query the trails. UbiTrail [20] uses an ontology to standardize the trails storage. Furthermore, the model has a mechanism that supports the creation of query

Fig. 1. The TrailTrade architecture.

J.L.V. Barbosa et al. / Computers in Industry 80 (2016) 43–53

45

approach to support trade in goods and services without domain restrictions. In general, related works explore some kind of context information. This literature revision did not find any research work addressing trail-aware in commerce.

through the Personal Assistant interface. The profile is stored in the environment server (Profile System) and also in the dealer’s mobile device. Each time the user logs in, the mobile device synchronizes with the server; so that dealers could change the environment, without losing their profile information.

3. TrailTrade model

3.3. Trails management

TrailTrade consists of seven components (see Fig. 1). The Location System allows the identification of users’ current location. The Profile System stores users’ basic information. The Trails Management manages the dealers’ trails. This component is implemented with the UbiTrail [20]. The Reference System determines the reputation of users according to previous business transactions. The Opportunity Consultant is the main component and works as an analysis engine to generate business opportunities. The Service Manager acts as a communication layer between the interface component and other components. These first six components form an environment server that can map regions and support multiple users. The Personal Assistant runs on mobile devices to support the interface with users. The next subsections detail each component.

The Trails Management is based on UbiTrail [20], which in turn is composed of a server and a client. The server supports the trails management and provides services to applications. The client runs on mobile devices and supports communication between applications and the server. A trail consists of a sequence of records in a data structure called ptrail. This structure is standardized by UbiTrail ontology [20], which was extended from the UbisWorld [53]. The trailpoint makes the trail composition. This process sends a ptrail from a mobile device to the server. The trailpoint occurs when the application automatically identifies an user has performed an important event, for example, entered or left a region, interacted with another user (for example, a completed deal), among others. Fig. 2 illustrates the trail composition of a mobile device user. Each star is the occurrence of an event which generates a trailpoint. UbiTrail provides basic services to manage trails. These services perform generic tasks which are considered strategic for trails management, such as trails composition and queries related to the trails. In addition, the server allows applications to register specialized services according to the needs. Section 3.6 describes four specialized services designed to support the discovery of deal opportunities driven by trails. The Trails Management interacts with two other components. The Opportunity Consultant calls specialized services to explore deal opportunities using the dealers’ trails, as discussed in Section 3.6. In turn, the Service Manager supports the interface between Personal Assistant and UbiTrail. The Assistant sends ptrails to UbiTrail through the Trail services of the Service Manager. The interaction between these three components will be better discussed in Sections 3.5,3.6 and 3.7.

3.1. Location system The Location System converts to regions the physical locations obtained by antennas triangulation (indoor) or GPS coordinates (outdoor). This component uses a location-aware system [50] which supports different techniques to determine the position of mobile devices. The technique chosen depends on available resources on the environment and mobile device. The process is as follows: (1) the Personal Assistant sends the physical location for the Location System using the Location service of Service Manager; (2) the Location System maps the physical location for a region; (3) the information of the region is provided to the Opportunity Consultant. 3.2. Profile system Research on users profiles has been applied to improve contextaware and pervasive/ubiquitous systems [14–16,18]. Typically, the studies focus on the use of meta-models and ontologies to standardize and organize the profile information [14,22,24] and strategies for profile management [16]. However, there is no standard profile for context-aware or pervasive/ubiquitous commerce systems. Some initiatives focus on e-commerce, mainly representing products related to intellectual property rights, such as books, music, DVDs and software [51,52]. In addition, a recent trend in adaptive systems [13] is to base decisions on users’ trails [16,17], rather than just on profiles. TrailTrade stores users’ basic information in profiles and uses their trails to improve the inference of business opportunities. TrailTrade organizes the profiles in a meta-model composed of four categories (see Table 1). Dealers provide profile information

3.4. Reference system Trust between dealers is important to do business, especially in e-commerce. Trust is even more critical when deals are encouraged through the use of contexts and trails because contact between dealers is recommended automatically. Reputation systems [54] help customers to decide whether suppliers are reliable, without knowing them directly. In addition, these systems encourage the good behavior by suppliers, because they expect their customers consider past behavior to achieve future business [55]. Reputation is crucial for context-aware and trail-aware commerce systems. So every time a deal takes place in TrailTrade, the dealers rank their counter-parties. The dealers’ behavior is stored in the trails, making possible to obtain detailed information

Table 1 Meta-model of profiles. Category

Description

Identification General data to identify the dealer Desires Products or services that the dealer is interested in acquiring Offers Products or services that the dealer is interested in supplying References Risk information about the dealer, as discussed in Section 3.4

Elements Company registry or social security number, name, address, telephone and e-mail Type of business following the UbiTrail ontology [20], title, characteristics, and payment terms Type of business following the UbiTrail ontology [20], title, characteristics, and payment terms Reference Grade of Dealer (RGD) and Degree of Tolerated Risk (DTR)

46

J.L.V. Barbosa et al. / Computers in Industry 80 (2016) 43–53

Fig. 2. Example of trail composion.

about their reputations. The Reference System uses a mechanism that calculates the average of evaluations [55,56]. The result is called Reference Grade of Dealer (RGD). This component updates the RGD in the dealer’s profile (Profile System) in each business transaction. When the Opportunity Consultant identifies a business opportunity, the Degree of Tolerated Risk (DTR) of dealers and their RGD are compared. The DTR is also maintained by the Profile System (see Table 1). The opportunity will only be ratified if the DTR is less than or equal to RGD. Sophisticated reputation services can be supported by the Service Manager using trails, such as dealers’ RGD in specific regions or their RGD in specific types of deals. The global RGD is calculated by Reference System and kept in the profiles as basic information to efficiently guide the deals. 3.5. Service manager The Service Manager operates as a communication layer between the components, mainly supporting the Personal Assistant. The services are organized into categories (Table 2) and follow types of coverage:  Environment: the services are available to all dealers anywhere in the TrailTrade coverage area;  Region: services available only to dealers that are within a region. For example, the service to book a meeting room at a trade fair can be provided only to dealers that are within a specific region;  Dealer: the services are available only to counterparts of a dealer in a negotiation. For example, the service product demonstration that a dealer offers to its customers;  Dealer within a region: the services are available only to counterparts of a dealer in a specific region. For example, Table booking service of a restaurant (dealer) can be accessed only when the client is inside the food court (region).

3.6. Opportunity consultant The Opportunity Consultant identifies opportunities by analysing profiles and trails. This component acts as recommender systems [58,59], which are widely used in e-commerce to increase the likelihood of customers to find products that meet their needs. The analysis in a region starts every time occurs: (1) a change of dealers in the region (Location System); (2) a change in their desires, offers or references (Profile System); (3) a change in their trails, such as when new opportunities or concluded deals are recorded in a trail (Trails Management). The Consultant identifies three types of opportunities: (1) Buy and Sell, when desires match offers; (2) Knowledge interchange, when desires match desires, identifying common tastes; (3) Experience interchange, when desires match completed deals recorded in trails of other dealers, thus identifying a possible exchange or sale of experiences. In addition, the Consultant can perform a more advanced discovery of opportunities through the specialized services (see category Services in Table 2), exploring trails according to the applications interests. This article proposes four specialized services to explore the trails (Table 3), which were implemented (Section 4) and used in the evaluation (Section 5). Fig. 3 summarizes the discovery of deal opportunities. The Location System determines in which region the dealers are (step 1). In addition, the Profile System obtains the dealers’ profiles (offers, desires, RGD and DTR). The Consultant can use the specialized services to explore trails, improving the search of opportunities (step 2). The Consultant determines risk compatibilities through RGD/DTR. If dealers are compatible and are involved in an opportunity, the Consultant sends a notification to them, using the Warnings service of the Service Manager (step 3). After, the Personal Assistant shows a summary of the opportunity and offers three options: store, dispose or view. The first two allow managing the opportunities for later use or disposal. The last

Table 2 TrailTrade services. Category

Description

Warnings Location

These services allow sending warnings to dealers about the opportunities. In addition, the dealers can used them to send messages to other dealers This service receives the physical location sent by the Personal Assistant and returns a symbolic region mapped in the TrailTrade’s coverage. The conversion is performed by the Location System Through these services the Personal Assistant accesses the Trails Management (UbiTrail [20]). The main service is to send the ptrail to create the trails These services allow dealers to submit proposals and counter-proposals for negotiation and also complete their deals. TrailTrade does not support payment. This can be supported by specific services included in the category Services Services in this category allow dealers to manage their profiles and also make the synchronization between the server and the mobile devices Dealers can include and find services in regions where they are. An example of non-standard service is the mobile payment (m-payment) [57]. For example, a restaurant could register this service, configuring it to the restaurant region. Consequently, the mobile payment would only be available to restaurant clients

Trail Dealing Profile Services

J.L.V. Barbosa et al. / Computers in Industry 80 (2016) 43–53

47

Table 3 Specialized services included in TrailTrade. Service

Description

This service returns a list of dealer’s preferred regions taking into account the number of times that the dealer visited the regions This service returns a list of dealer’s preferred regions to deal, considering the regions where the dealer concluded more deals This service returns a list of opportunities in a region, considering the dealers in the region and that they have: (1) the region as preferred to complete deals (Preferred_Dealing_Regions); OR (2) compatibility among desires and offers in their profiles; AND (3) risk compatibility considering their RGD and DTR (profiles). The service considers the first two aspects as complementary, but the third as mandatory. Even if dealers do not have desires and offers in common, they may like to deal in a region, and this can be used to foster deals of products and services not mentioned in their profiles Type_of_Business_by_Region This service returns a list of preferred deals types by region. This service considers all completed deals in the regions. This information can be used to promote a specific type of deal in a region, for example, some neighborhoods or streets have a business profile (as Chinatown to deal Chinese products), which can be informed to visitors Preferred_Regions Preferred_Dealing_Regions Discovery_DOs_by_Region

shows a detailed description of the opportunity containing the product or service characteristics, the amount involved in the deal and the payment terms. Section 5.3 discusses a scenario and shows snapshots of the Assistant (see Fig. 7 in advance).

3.7. Personal assistant The Personal Assistant runs on mobile devices and interacts with the Service Manager to: (1) identify the dealer and request access (login); (2) read the physical location and send to Location System;

Fig. 3. Discovery of deal opportunities.

Fig. 4. Map of the simulation environment.

48

J.L.V. Barbosa et al. / Computers in Industry 80 (2016) 43–53

(3) temporarily store the profiles and trails to synchronize with the server; (4) receive and display warnings; (5) allow dealers to manage their profiles. 4. Implementation issues Implementation used the C# language. The Personal Assistant captures the signal strength of the antennas via the OpenNETCF library and forwards the information to the Location System through the Location service. TrailTrade calculates where the mobile device is, through an algorithm that converts the intensity of the signals to regions [50]. The validation environment used four antennas as discussed in Section 5.2. The Profile System uses the relational database Firebird and profiles on mobile devices are stored by .NET System.Xml Library. The Opportunity Consultant and the Reference System were implemented in C# as Windows services. The Services Manager uses webservices technology and Trails Management is based on UbiTrail prototype [20] which was implemented using Java language, MySQL and webservices. 5. Application and results The scientific community has used scenarios to assess contextaware systems (see Dey et al. [10] and Barbosa et al. [23]), ubiquitous systems (Satyanarayanan [48] and recent works [14,22,24]), trail-aware systems [16,17,20] and ubiquitous commerce systems [24]. Following this strategy, a commerce scenario was applied to assess the TrailTrade. The scenario was motivated by the train that connects two cities located in southern Brazil, called Porto Alegre and Novo Hamburgo. This train passes near two universities, two faculties, three malls and daily carries thousands of people. A scenario with this diversity of regions and spread over a wide geographic area is strategic for the evaluation of trail-aware systems. The scenario can consider indoor regions (for example, stores and food court in a mall) and outdoor regions (paths between the houses, stations, train and mall). The technological complexity to implement an experiment involving real-life situations in this broad scenario stimulated the use of simulated trails (Section 5.1) and their application in a real environment based on the scenario (Section 5.2). We understand the real environment as the building where the infrastructure to support the experiment was installed. 5.1. Obtainment of trails Fig. 4 shows the simulated environment used to generate trails. The environment is composed of three regions: Houses, Metro and Shopping. Regions may be composed of sub-regions and each region contains at least one location. The Metro region maps the movement of dealers from their houses to the shopping and vice versa. This region contains locations for three stations and a location that represents the dealer traveling in the train. The Shopping region has two sub-regions which represent the stores and a food court.

Dealers’ profiles were generated considering the two subjects who participated in the execution of the scenario described in Section 5.3. The profiles recorded preferences and interests related to desires and offers. The trails were generated using an application developed specifically for this purpose. The records were related to events of dealers’ day-to-day, such as entry and exit from locations, inclusion of offers and desires, opportunities detected and concluded businesses. Table 4 shows examples of these records. 5.2. Validation environment Fig. 5 shows the map of the second floor of the building where the validation environment was implemented. The environment consists of nine rooms, in which were installed four wireless antennas. The Location System was implemented through antennas triangulation [50] and allowed the precise determination of the rooms where users were. The rooms were mapped to the simulation environment shown in Fig. 4. Therefore, a relationship was established between the real environment (Fig. 5) and the simulated environment (Fig. 4) in accordance with that shown in Fig. 6. Rooms 206 and 215 were mapped to locations Train and 1 dealer’s house. The other rooms were mapped to the Shopping region. Movements between real rooms were equivalent to movements in the simulated environment. For example, a movement from room 215 (house) to room 206 (train) represented the dealer traveling from the house to the shopping. When the dealer entered in room 206 (train), Opportunity Consultant started the analysis process of deal opportunities, as if the dealer had actually boarded the train. 5.3. Validation through a business scenario TrailTrade was installed in the evaluation environment and two volunteers carrying mobile devices moved through the rooms where they acted and interacted following a scripted scenario. The scenario involved the daily routine of a jewelry appraiser, called Peter, and a music student named Eduard as described in the following paragraphs. “Peter is a jewelry appraiser in a store in the Shopping region. Earlier in the day, before leaving home, he remembers that he would like to sell his old guitar, which he does not use anymore. So, he records an offer of the guitar in the Profile System. After that, the Opportunity Consultant looks for deals opportunities involving guitar each time the user enters a new region. In this scenario the locations were also mapped as regions. The Consultant uses the Discovery_DOs_by_Region service to look for opportunities considering preferred regions for deals (Preferred_Dealing_Regions service) or any particular desire related to a guitar in any region. In both cases, the risk compatibility of dealers is respected. To arrive at work, Peter goes to Station 1 and takes the Train. He usually uses the travel interval to check his messages and deals through the Personal Assistant. Arriving at the Shopping region he works until 12:00 am. Then he goes to the Fast food 1 region for lunch, where he remains until 12:45. After lunch,

Table 4 Example of trails records. Dealer

Event

Location

Extension

Date

Peter

Offer inclusion

House

February. 6 08:03

Peter Peter Eduard Eduard Eduard

Exit from location Entry into location Entry into location Entry into location Created Opportunity

House Fast Food 1 Shopping Fast Food 1 Fast Food 1

Title: Guitar Category: Sound Instrument – – – – DO Notification

February. February. February. February. February.

6 6 6 6 6

08:11 12:01 15:32 16:03 16:04

J.L.V. Barbosa et al. / Computers in Industry 80 (2016) 43–53

49

Fig. 5. Map of the real validation environment.

Fig. 6. Real environment mapped to the simulation environment.

he returns to the Jewelry store. Finally, at the end of the day, Peter moves to the Shopping station, and goes home. Eduard is a music student and wants to buy a guitar, so he records this desire in his profile. Every morning, he goes to Station 2 and takes the train to the university. The university region is not mapped in the scenario. After the class, in the afternoon, he takes the train and goes to the Shopping region to meet with his friends. After the meeting, he moves to the Shopping station to go home. Opportunity Consultant uses the Preferred_Regions service, which takes in account the frequency in which regions are visited, to determine the preferred regions of each user. In this scenario the Jewelry store and the Fast food 1 were considered the preferred regions of Peter. In a regular day, Eduard takes the train to the university, has classes and goes to the Shopping region to meet his friends. Sometimes, they go to Fast food 1 in the Food court region. In a specific time, Eduard and his friends go to the same place again (Fast food 1). The Opportunity Consultant identifies for the first time through the Preferred_Regions service, Fast food 1 as a preferred region of Eduard. This occurs because the service considers a number of entrances to determine preferred regions, and in this time is reached the limit to this indication. The Opportunity Consultant uses the Preferred_Regions service to get the dealers who have the Fast food 1 as preferred region. As Peter and Eduard usually pass through the Fast food 1 (Preferred region) and have compatibilities between offer/desire (the guitar) and RGD/DTR, for them would be interesting to make a schedule in this place to talk about the business. We believe that places visited by both dealers can serve to hold meetings and even for delivery of products.

Therefore, the Consultant sends a message to Eduard (Fig. 7a). Thus, he is able to exchange messages with the counterpart, in this case, Peter. Whenever the dealers receive a notification, they can delay access. This action indicates that they are interested in the deal but do not want to access it at the time. In the scenario, Eduard postpones the notification (Fig. 7b), because he is in a meeting with his friends. After the meeting, Eduard moves to the Shopping station. When he enters the train, TrailTrade notifies him that a postponed opportunity was available (Fig. 7c). Eduard had indicated the train to read the postponed messages. He then accesses the opportunity (Fig. 7d) and exchanges messages with Peter. They reserve a day and time in the Fast food 1 at the Shopping region to carry out the deal.” 5.4. Discussion on validation The objective of the case study was to evaluate the TrailTrade’s functionalities, mainly its trail awareness support. In this sense, the results were encouraging once the trails were used to foster a negotiation based on past behaviour of dealers. In the scenario, the deal would not occur without the analysis of the preferred regions. Despite the dealers have never been in a region at the same time, they used the same preferred region to articulate the business. We believe the validation through the scenario proved the following relevant aspects of trail-aware commerce: (1) the feasibility of trails usage to articulate and improve deals; (2) the context-aware and pervasive/ubiquitous systems as strategic technologies to promote deals; (3) the potential use of trails to improve automatic inference of deals in context-aware and pervasive/ubiquitous commerce environments. Additionally, functionalities tested in the

50

J.L.V. Barbosa et al. / Computers in Industry 80 (2016) 43–53

Fig. 7. The scenario snapshots.

scenario worked adequately, such as: (1) trails management using UbiTrail [20]; (2) indoor location system based on antennas triangulation [50]; (3) profile and reference systems to manage the risk information of dealers; (4) personal assistant running on mobile devices and; (5) mainly, specialized services to trail awareness, emphasizing the Preferred_Regions service. As discussed in Section 5.1, the scenario was based on trails generated through simulation for only two users. Trails were stored through the trails management system called UbiTrail [20]. During the execution, TrailTrade consulted UbiTrail to get information of trails and implement the specialized services (Table 3). Moreover, the execution was conducted in a validation environment specifically prepared to the scenario (Section 5.2). Therefore, the number of users and the controlled environment did not allow the assessment of TrailTrade performance when submitted to massive number of users and real-life environments. The authors intend to expand the current evaluation for a largescale test with a focus on smart cities [60]. It can be argued that the generic and broad approach of TrailTrade can be used to implement this idea. The large-scale scenario is more complex

and will introduce restrictions, especially related to the overhead: (1) in storage and response time of trails management services and (2) in the massive processing of specialized services to trail awareness. In this sense, quantitative evaluations would be strategic, such as the evaluation of performance and scalability. In addition, the focus on functionalities and the small number of volunteers could not provide usability feedback from users. TrailTrade’s usability is a strategic issue to be evaluated as will be addressed in the next section. 6. Conclusion This article introduced a model for trail-aware commerce support, called TrailTrade. The model was implemented and evaluated through a scenario. The scenario made possible the execution of a prototype in a real validation environment, even using simulated trails. The literature review showed that TrailTrade is the first model that uses dealers’ trails to promote business opportunities. In addition, the prototype and the experiment conducted in a real

J.L.V. Barbosa et al. / Computers in Industry 80 (2016) 43–53

environment attested to the proposal feasibility, encouraging the application of TrailTrade in real business situations. TrailTrade model consists of an initial proposal as well as the current prototype. According to the results and limitations perceived during the research, some proposals for new studies were identified. There is room to expand the evaluation. Additional tests with simulated trails will be conducted. Type_of_Business_by_Region service was not used, and Preferred_Dealing_Regions and Discovery_DOs_by_Region services were executed, but their potential to promote deals have not been exploited in the scenario. The heuristic to determine the preferred regions (visit frequency) can be improved by additional information (e.g., the time of each visit), thereby improving the Preferred_Regions service. Moreover, the Reference System does not take into account the different profiles of dealers who made the rankings. This component could be improved, making it able to use Collaborative Filtering [61] to calculate the reputation based on the assessments made by dealers with similar profiles. Moreover, the following future developments could improve the TrailTrade: (1) new specialized services could be implemented to perform additional discovery of opportunities, increasing the usefulness of the Opportunity Consultant; (2) trails could be used to implement profile management [16], allowing the use of dynamic profiles instead of static profiles currently in use; (3) the Opportunity Consultant could use context prediction based on trails [45] to infer potential deals, improving the trail awareness with considerations about the future. Finally, the work could be enriched with an evaluation of the technology acceptance. A possible strategy would be to run the scenario described in Section 5.3 with a larger number of individuals. Another would be to use a more complete scenario involving all specialized services (Table 3) in potential real-life situations, such as: (1) Preferred_Dealing_Regions to limit the notification of opportunities only to the regions and times preferred, for example, only when the user is having lunch in the food court of a mall; (2) Discovery_DOs_by_Region to promote deals between people who are participating in a trade fair (wide area), or focusing on a meeting room (most restricted region); (3) Type_of_Business_by_Region to use the specific kind of deal being done in a fair booth to promote more negotiations on it. After the execution, the users would answer an assessment questionnaire based on the Technology Acceptance Model (TAM). The TAM has been considered a standard to evaluate the acceptance of new technologies [62]. The model considers as the main influences for the acceptance:  Perceived ease of use: the degree of how many people believe that the technology will reduce their efforts;  Perceived usefulness: the degree of how many people believe that the technology could improve their performance. The questionnaire would consist of statements to measure the impact of the trail awareness in commerce support, such as:  Perceived ease of use: (1) the services based on trails are easy to understand and use; (2) the opportunities created through trails were presented in a clear and simple manner; (3) the trails information were treated transparently;  Perceived usefulness: (1) the services based on trails promote the identification of opportunities; (2) the opportunities created were in accordance with my desires; (3) TrailTrade is useful to stimulate interaction with other dealers; (4) the system will be commercially useful. The participants would fill out the questionnaire following the Likert scale of five points [63], spanning from 1 (completely

51

disagree) to 5 (completely agree). The results will be depicted in two graphs to summarize the results in each evaluation factor. Acknowledgments The authors wish to acknowledge that this work was supported by CNPq/Brazil (National Council for Scientific and Technological Development—http://www.cnpq.br) and Capes/Brazil (Coordination for the Improvement of Higher Education Personnel—http:// www.capes.gov.br). We are also grateful to Unisinos (http://www. unisinos.br) and Feevale (http://www.feevale.br) for embracing this research. Finally, we would like to express our sincere thanks to the reviewers for their valuable contributions to the final quality of this article. References [1] A. Diaz, P. Merino, F.J. Rivas, Mobile application profiling for connected mobile devices, IEEE Pervasive Comput. 9 (1) (2010) 54–61, doi:http://dx.doi.org/ 10.1109/mprv.2009.63. [2] M. Satyanarayanan, P. Bahl, P. Caceres, N. Davies, The case for VM-based cloudlets in mobile computing, IEEE Pervasive Comput. 8 (4) (2009) 14–23, doi:http://dx.doi.org/10.1109/mprv.2009.82. [3] X. Chen, S. Lian, Service and P2P based secure media sharing in mobile commerce environments, Electron. Commer. Res. 11 (1) (2011) 91–101, doi: http://dx.doi.org/10.1007/s10660-010-9069-6. [4] J. Hightower, G. Borriello, Location systems for ubiquitous computing, Computer 34 (8) (2001) 57–66, doi:http://dx.doi.org/10.1109/2.940014. [5] J. Hightower, A. LaMarca, I. Smith, Practical lessons from place lab, IEEE Pervasive Comput. 5 (3) (2006) 32–39, doi:http://dx.doi.org/10.1109/ mprv.2006.55. [6] A. Dey, J. Hightower, E. Lara, N. Davies, Location-based services, IEEE Pervasive Comput. 9 (1) (2010) 11–12, doi:http://dx.doi.org/10.1109/mprv.2010.10. [7] S.J. Vaughan-Nichols, Will mobile computing’s future be location? Computer 42 (2) (2009) 14–17, doi:http://dx.doi.org/10.1109/MC.2009.65. [8] K. Petrova, B. Wang, Location-based services deployment and demand: a roadmap model, Electron. Commer. Res. 11 (1) (2011) 5–29, doi:http://dx.doi. org/10.1007/s10660-010-9068-7. [9] A. Aloudat, K. Michael, Toward the regulation of ubiquitous mobile government: a case study on location-based emergency services in Australia, Electron. Commer. Res. 11 (1) (2011) 31–74, doi:http://dx.doi.org/10.1007/ s10660-010-9070-0. [10] A. Dey, D. Salber, G. Abowd, A conceptual framework and a toolkit for supporting the rapid prototyping of context-aware application, Hum.– Comput. Interact. 16 (2) (2001) 97–166, doi:http://dx.doi.org/10.1207/ S15327051HCI16234_02. [11] M. Knappmeyer, S.L. Kiani, E.S. Reetz, N. Baker, R. Tonjes, Survey of context provisioning middleware, IEEE Commun. Surv. Tutorials 15 (3) (2013) 1492– 1519, doi:http://dx.doi.org/10.1109/surv.2013.010413.00207. [12] P. Bellavista, A. Corradi, M. Fanelli, L. Foschini, A survey of context data distribution for mobile ubiquitous systems, ACM Comput. Surv. 44 (4) (2012) 1–45, doi:http://dx.doi.org/10.1145/2333112.2333119. [13] J. Lopes, R. Souza, C. Geyer, C.A. Costa, J. Barbosa, M. Gusmão, A.C. Yamin, Managing adaptation in ubicomp, Proceedings of the Latin Conference of Informatics (CLEI), Medellin, 2012, pp. 1–8, doi:http://dx.doi.org/10.1109/ CLEI.2012.6427254. [14] J.L.V. Barbosa, R.M. Hahn, D.N.F. Barbosa, A.I.C.Z. Saccol, A ubiquitous learning model focused on learner interaction, Int. J. Learn. Technol. 6 (1) (2011) 62–83, doi:http://dx.doi.org/10.1504/IJLT.2011.040150. [15] J.L.V. Barbosa, D.N.F. Barbosa, J.M. Oliveira, S.A.J. Rabello, A decentralized infrastructure for ubiquitous learning environments, J. Univ. Comput. Sci. 20 (2) (2014) 1649–1669, doi:http://dx.doi.org/10.3217/jucs-020-12-1649. [16] A. Wagner, J.L.V. Barbosa, D.N.F. Barbosa, A model for profile management applied to ubiquitous learning environments, Expert Syst. Appl. (2013) 1–12, doi:http://dx.doi.org/10.1016/j.eswa.2013.08.098. [17] R.R. Oliveira, I.G. Cardoso, J.L.V. Barbosa, C.A. Costa, & Prado, An intelligent model for logistics management based on geofencing algorithms and RFID technology, Expert Syst. Appl. (2015), doi:http://dx.doi.org/10.1016/j. eswa.2015.04.001. [18] J. Hong, E. Suh, J. Kim, S. Kim, Context-aware system for proactive personalized service based on context history, Expert Syst. Appl. 36 (4) (2009) 7448–7457, doi:http://dx.doi.org/10.1016/j.eswa.2008.09.002. [19] C. Driver, S. Clarke, An application framework for mobile, context-aware trails, Pervasive Mob. Comput. 4 (5) (2008) 719–736, doi:http://dx.doi.org/10.1016/j. pmcj.2008.04.009. [20] J.M. Silva, J.H. Rosa, J.L.V. Barbosa, D.N.F. Barbosa, L.M. Palazzo, Content distribution in trail-aware environments, J. Braz. Comput. Soc. 16 (3) (2010) 163–176, doi:http://dx.doi.org/10.1007/s13173-010-0015-1. [21] Wagner Cambruzzi, S.J. Rigo, J.L.V. Barbosa, Dropout prediction and reduction in distance education courses with the learning analytics multitrail approach,

52

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31] [32]

[33]

[34] [35]

[36]

[37]

[38]

[39]

[40] [41]

[42]

[43]

[44]

[45]

[46]

J.L.V. Barbosa et al. / Computers in Industry 80 (2016) 43–53 J. Univ. Comput. Sci. 21 (1) (2015) 23–47, doi:http://dx.doi.org/10.3217/jucs020-12-1649. H.D. Vianna, J.L.V.A. Barbosa, Model for ubiquitous care of noncommunicable diseases, IEEE J. Biomed. Health. Inf. 18 (5) (2014) 1597–1606, doi:http://dx.doi. org/10.1109/JBHI.2013.2292860. J.L.V. Barbosa, M.R. Kich, D.F.B. Barbosa, A.Z. Klein, S.J. Rigo, DeCom: a model for context-aware competence management, Comput. Ind. 72 (2015) 27–35, doi: http://dx.doi.org/10.1016/j.compind.2015.03.012. L.K. Franco, J.H. Rosa, J.L.V. Barbosa, C.A. Costa, A.C. Yamin, MUCS: a model for ubiquitous commerce support, Electron. Commer. Res. Appl. 10 (2) (2011) 237– 246, doi:http://dx.doi.org/10.1016/j.elerap.2010.08.006. P. Benou, C. Vassilakis, The conceptual model of context for mobile commerce applications, Electron. Commer. Res. 10 (2) (2010) 139–165, doi:http://dx.doi. org/10.1007/s10660-010-9050-4. P. Eugster, B. Garbinato, A. Holzer, Pervaho: a specialized middleware for mobile context-aware applications, Electron. Commer. Res. 9 (4) (2009) 245– 268, doi:http://dx.doi.org/10.1007/s10660-009-9042-4. U. Panniello, M. Gorgoglione, Incorporating context into recommender systems: an empirical comparison of context-based approaches, Electron. Commer. Res. 12 (1) (2012) 1–30, doi:http://dx.doi.org/10.1007/s10660-0129087-7. K. Lin, T. Yu, C. Chith, The design of a personal and intelligent pervasive commerce system architecture, Proceedings of the 2nd IEEE International Workshop on Mobile Commerce and Services, Munich, 2005, pp. 163–173, doi: http://dx.doi.org/10.1109/WMCS.2005.25. D. Mazzari, P. Seraphine, S. Ahamed, S. Balasubramanian, Design and implementation of EPARK wireless parking payment system using pervasive computing, Proceedings of the 2nd IEEE International Symposium on Pervasive Computing and Ad Hoc Communications, Niagara Falls, 2007, pp. 219–224, doi:http://dx.doi.org/10.1109/AINAW.2007.145. S. Shekar, P. Nair, S. Helal, iGrocer: a ubiquitous and pervasive smart grocery, Proocedings of the ACM Symposium on Applied Computing, Melbourne, 2003, pp. 645–652, doi:http://dx.doi.org/10.1145/952532.952658. G. Roussos, Ubiquitous and Pervasive Commerce: New Frontiers for Electronic Business, Springer, New York, 2006. L. Zhang, Q. Liu, X. Li, Ubiquitous commerce: theories, technologies, and applications, J. Networks 4 (4) (2009) 271–278, doi:http://dx.doi.org/10.4304/ jnw.4.4. G. Roussos, T. Moussouri, Consumer perceptions of privacy, security and trust in ubiquitous commerce, Pers. Ubiquitous Comput. 8 (6) (2004) 416–429, doi: http://dx.doi.org/10.1007/s00779-004-0307-6. R.T. Watson, U-commerce: the ultimate, Mag. Ubiquity 2000 (2000), doi: http://dx.doi.org/10.1145/353165.353882. H. Galanxhi-Janaqi, F.F. Nah, U-commerce: emerging trends and research issues, Ind. Manage. Data Syst. 104 (9) (2004) 744–755, doi:http://dx.doi.org/ 10.1108/02635570410567739. G. Roussos, A. Gershman, P. Kourouthanassis, Ubiquitous commerce, Proceedings of the Fifth International Conference on Ubiquitous Computing— Ubiquitous Commerce Workshop, Seattle, 2003, pp. 1–4. A. Gershman, Ubiquitous commerce: always on, always aware, always proactive, Proceedings of the Symposium on Applications and the Internet, Nara, 2002, pp. 37–38, doi:http://dx.doi.org/10.1109/SAINT.2002.994443. D. Chiu, H. Leung, Towards ubiquitous tourist service coordination and integration: a multi-agent and semantic web approach, Proceedings of the International Conference on Electronic Commerce (ICEC), New York, 2005, pp. 574–581, doi:http://dx.doi.org/10.1145/1089551.1089656. A. Ciaramella, M. Cimino, B. Lazzerini, F. Marcelloni, Using Context history to personalize a resource recommender via a genetic algorithm, Proceedings of the International Conference on Intelligent Systems Design and Applications (ISDA), Cairo, 2010, pp. 965–970, doi:http://dx.doi.org/10.1109/ ISDA.2010.5687064. V. Bush, As we may think, Atlantic Mon. 176 (1) (1945) 101–108. S. Ashley, (2008). Who Controls the Past Controls the Future—Life Annotation in Principle and Practice. University of Southampton, School of Electronics and Computer Science, Doctoral Thesis. http://eprints.soton.ac.uk/266554/1. hasCoversheetVersion/thesis.pdf (accessed June 2015). A.R. Doherty, N. Caprani, C. Conaire, V. Kalnikaite, C. Gurrin, A.F. Smeaton, N.E. O’Connor, Passively recognising human activities through lifelogging, Comput. Hum. Behav. 27 (5) (2011) 1948–1958, doi:http://dx.doi.org/10.1016/j. chb.2011.05.002. A.J. Sellen, S. Whittaker, Beyond total capture: a constructive critique of lifelogging, Commun. ACM 53 (5) (2010) 70–77, doi:http://dx.doi.org/10.1145/ 1735223.1735243. N. Gyöbíró, Á. Fábián, G. Hományi, An activity recognition system for mobile phones, Mob. Networks Appl. 14 (1) (2009) 82–91, doi:http://dx.doi.org/ 10.1007/s11036-008-0112-y. J.H. Rosa, J.L.V. Barbosa, G.O. Barcelos, ORACON: an adaptive model for context prediction, Expert Syst. Appl. 45 (1) (2016) 56–70, doi:http://dx.doi.org/ 10.1016/j.eswa.2015.09.016. D. Manikandan, J. Madhusudhanan, V.V. Prasanna, V. Amtirh, M.A. Britto, Smart banking environment based on context history, Proceedings of the International Conference on Recent Trends in Information Technology (ICRTIT), Chennai, 2011, pp. 450–455, doi:http://dx.doi.org/10.1109/ ICRTIT.2011.5972335.

[47] R. Caceres, A. Friday, Ubicomp systems at 20: progress, opportunities, and challenges, IEEE Pervasive Comput. 11 (1) (2013) 14–21, doi:http://dx.doi.org/ 10.1109/MPRV.2011.85. [48] M. Satyanarayanan, Pervasive computing: vision and challenges, IEEE Pers. Commun. 8 (4) (2001) 10–17, doi:http://dx.doi.org/10.1109/98.943998. [49] M. Satyanarayanan, A catalyst for mobile and ubiquitous computing, IEEE Pervasive Comput. 1 (1) (2002) 2–5, doi:http://dx.doi.org/10.1109/ MPRV.2002.993138. [50] C. Rolim, N. Sonntag, J.L.V. Barbosa, HLS: a model for developing locationaware applications, Proceedings of the 11th Workshop High Performance Computational Systems, Campo Grande, 2008, pp. 227–234. [51] G. Rust, M. Bide, (2000). The metadata framework: Principles, model and data dictionary. Technical Report WP1a-006-2.0. http://www.doi.org/ topics/indecs/indecs_framework_2000.pdf (accessed June 2015). [52] N. Medeiros, Metadata for e-commerce: the ONIX International standard, OCLC Syst. Serv. 17 (3) (2001) 114–117, doi:http://dx.doi.org/10.1108/ EUM0000000005816. [53] D. Heckmann, M. Loskyll, R. Math, P. Recktenwald, C. Stahl, UbisWorld 3.0: a semantic tool set for ubiquitous user modeling, Proceedings of the International Conference on User Modeling, Adaptation, and Personalization (UMAP), Trento, 2009, pp. 1–6. [54] P. Resnick, K. Kuwabara, R. Zeckhauser, E. Friedman, Reputation systems, Commun. ACM 43 (12) (2000) 45–48, doi:http://dx.doi.org/10.1145/ 355112.355122. [55] P. Resnick, R. Zeckhauser, Trust among strangers in internet transactions: empirical analysis of eBay’s reputation system, in: M.R. Baye (Ed.), The Economics of the Internet and E-Commerce, Elsevier, Netherlands, 2002, pp. 127–157. [56] A. JØsang, R. Ismail, C. Boyd, A survey of trust and reputation systems for online service provision, Decis. Support Syst. 43 (2) (2007) 618–644, doi:http://dx. doi.org/10.1016/j.dss.2005.05.019. [57] A. Roehrs, C.A. Costa, J.L.V. Barbosa, A proposal of a mobile payment system based on android, Proceeding of the 6th International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing (IMIS), Palermo, 2012, pp. 617–622, doi:http://dx.doi.org/10.1109/IMIS.2012.86. [58] S. Weng, B. Lin, W. Chen, Using contextual information and multidimensional approach for recommendation, Expert Syst. Appl. 36 (2) (2009) 1268–1279, doi:http://dx.doi.org/10.1016/j.eswa.2007.11.056. [59] J. Bobadilla, F. Ortega, A. Hernando, A. Gutiérrez, Recommender systems survey, Knowl.-Based Syst. 46 (2013) 109–132, doi:http://dx.doi.org/10.1016/j. knosys.2013.03.012. [60] S. Choon, G. Byoung, Taxonomy of ubiquitous computing service for city development, Pers. Ubiquitous Comput. 17 (7) (2016) 1475–1483, doi:http:// dx.doi.org/10.1007/s00779-012-0583-5. [61] M. Zuo, J. Li, G. Liu, An adaptive collaborative filtering algorithm for online reputation systems, Proceedings of the 3rd International IEEE Conference on Signal-Image Technologies and Internet-Based System, Shanghai, 2007, pp. 1029–1035, doi:http://dx.doi.org/10.1109/SITIS.2007.72. [62] N. Maranguni c, A. Grani c, Technology acceptance model: a literature review from 1986 to 2013; Universal Access in the Information Society, Published online, 1–15, (2014), doi: 10.1007/s10209-014-0348-1. [63] R. Likert, A technique for the measurement of attitudes, Arch. Psychol. 22 (140) (1932) 1–55. Jorge Luis Victória Barbosa received the M.Sc. and Ph.D. degrees in Computer Science from the Federal University of Rio Grande do Sul, Porto Alegre, Brazil, in 1996 and 2002, respectively. He is currently a Full Professor of the Applied Computing Graduate Program (PIPCA) at the University of Vale do Rio dos Sinos (UNISINOS), São Leopoldo, Brazil. Additionally, he is a researcher of productivity at CNPq/Brazil and the head of the Mobile Computing Laboratory (MobiLab/UNISINOS). His research interests include mobile and ubiquitous computing, ubiquitous learning, ubiquitous commerce and ubiquitous accessibility. He is a member of the Brazilian Computer Society (SBC). Cládio José Martins Jr. is graduated in Computer Engineering at the University of Vale do Rio dos Sinos (UNISINOS), where he worked on the application of ubiquitous computing in business to generate deal opportunities at the Mobile Computing Laboratory (MobiLab). He also holds a Master in Electrical Engineering at Federal University of Santa Catarina (UFSC), where he worked in activities related to digital systems and wireless sensor networks in the Embedded Systems Group (GSE). He also teamed with the Part-TimeScientists (Google Lunar XPrize Group), where he participated in activities related to embedded systems and human interfaces. Nowadays he works as a Software Engineer in a Space Technology company in Munich, Germany. His research interests are mobile computing, location-based systems and distributed computing.

J.L.V. Barbosa et al. / Computers in Industry 80 (2016) 43–53

Laerte Kerber Franco received the M.Sc. degree in the Applied Computing Graduate Program (PIPCA) at the University of Vale do Rio dos Sinos (UNISINOS), São Leopoldo, Brazil, in 2009. He also received the MBA degree in Business Strategy from University of São Paulo (USP), Brazil, in 2015. He is currently manager of IT Governance area in a big construction company in São Paulo, Brazil. Additionally, he is a professor of Project Management, Total Quality Management and IT Governance at Mackenzie Presbyterian University and Faculty of Informatics and Management Paulista (FIAP) of São Paulo, Brazil. His research interests include mobile computing, ubiquitous computing, IT Governance and Business Management.

53

Débora Nice Ferrari Barbosa received the M.Sc. and Ph.D. degrees in Computer Science from the Federal University of Rio Grande do Sul, Brazil, in 2001 and 2007, respectively. Nowadays, she is a professor and researcher at the FEEVALE University, Novo Hamburgo, Brazil. Additionally, she is a researcher of productivity at CNPq (the Brazilian Council for Scientific and Technological Development). Her research interests include ubiquitous learning systems, distributed computing, multi-agent systems and artificial intelligence. She is a member of the Brazilian Computer Society (SBC).