Context-aware intelligent service system for coal mine industry

Computers in Industry 65 (2014) 291–305

Contents lists available at ScienceDirect

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

Context-aware intelligent service system for coal mine industry Xiao Xue *, Jing-kun Chang, Zhi-zhong Liu College of Computer Science, Henan Polytechnic University, Jiaozuo, China

A R T I C L E I N F O

A B S T R A C T

Article history: Received 2 December 2012 Received in revised form 19 June 2013 Accepted 12 November 2013 Available online 6 December 2013

In coal mine industry, the context-aware intelligent service system can be used to provide the most appropriate information services to miners according to their real-time situation. As a result, the information resources can be taken full advantage of so as to help miners to improve their safety condition, which are not just accessible to the management staff. However, there is not a detailed discussion or an in-depth research on the implementation of service system at present. In order to bridge the gap between the theory and the practice in the field, three critical problems need to be solved: (1) How to model the served miners’ context? (2) How to provide the information services to meet miners’ customized demands? (3) How to verify the availability of service invocation? According to the characteristics and practical needs of coal mine, this paper first proposes the Coal Mine Semantic Sensor Network Ontology (CMSSN) to build miner’s context model, which can facilitate reusing the context resources in different coal mines. Then, the configuration model connecting context information and business services is constructed to realize the customized service invocation. Thirdly, the method of computational experiment is proposed to verify the availability and validity of service system, i.e. whether the service system can provide the suitable service on time in various virtual accident experiments. A case study is given to explain how to implement the computational experiment on a virtual coal mine simulation platform. Finally, the potential problems in realizing the system are discussed, which will be our research focus in future. PACS: 75.40.-s; 71.20.LP ß 2013 Elsevier B.V. All rights reserved.

Keywords: Context-awareness Intelligent service system for mine industry Context-information-modeling Service composition Computational experiment

1. Introduction Information technology has been widely used in coal mine industry, especially in the fields of disaster prevention and emergency management. The seamless integration between computer technology and production management can effectively promote the operation of coal mine. In the past decade, a considerable progress has been achieved in the informatization of coal mine industry in China. However, there is still a large gap compared with other developed countries, such as the United States, Germany. According to the statistics of literature [1], in the past five years (2007–2011), China’s annual coal production is 3 times that of the United States (3 billion tons in China, 1 billion tons in the United States), but its mortality rate (every one million tons) is nearly 30 times that of the United States (China is 0.96, the United States is 0.03). The main causes for coal mine accidents are various. In addition to the natural conditions of the coal mine, mining technology, management level, and personnel quality, the application of information technology is considered to be a very

* Corresponding author. Tel.: +86 391 3989805. E-mail addresses: [email protected], [email protected] (X. Xue). 0166-3615/$ – see front matter ß 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.compind.2013.11.010

critical factor which account for the accidents in coal mine to a large extent. However, the development of digital mine has encountered many problems, some of which are as follows: (1) The number of information systems is growing continuously, but each of them is an isolated system with a special purpose. It is difficult to realize the interactions between two or more information systems, which result in data silos. As a result, how to take full advantage of the legacy systems to provide an integrated service has become a barrier to the development of mine industry. (2) The primary mode of delivering information services is PULL rather than PUSH. The management staffs are usually considered as target users who enjoy the information services, while the vast majority of underground miners feel it hard to enjoy the benefits of informatization. Therefore, another problem emerges concerning how to extend the scope of the information service so that they can be available for every miner. In order to provide an effective solution to the above problems, by inspiration of the idea of ‘‘smart planet’’ [2], the concept of ‘‘intelligent mine’’ [3] is put forward, which is expected to be more thorough perception (Instrumented), more comprehensive interconnected, and more intelligent than the digital mines by means of applying a lot of advanced technologies (such as service computing,

292

X. Xue et al. / Computers in Industry 65 (2014) 291–305

cloud computing, internet of things, etc.). Context-aware service is a key technology to build ‘‘intelligent mine’’, which is able to bridge the gap between the existing information services and miners’ demands. It can provide miners with the real-time services based on their own situation. However, the current research and application in this field are at the initial stage, which has posed a barrier for the implementation and promotion of intelligent mine. Literature [4] proposes an idea of how to apply context-aware technology in coal mine industry, and also presents the architecture of the context-aware-based service system. The architecture is composed of three layers: perceive the context information of service object (miners) by means of sensors; determine the current status of the served miners based on their context information; select and provide the appropriate business services. However, there is no an in-depth discussion of the implementation of service system in [4]. In order to promote the research in the field and facilitate the application of context-aware technology in coal mine industry, the following key issues need to be emphasized: (1) How to depict the miner’s context to support the subsequent service invocation; (2) How to provide the customized information service to miners in different situations, which can effectively deal with their particular needs and improve their safety condition; (3) How to verify the availability and validity of service system in order to facilitate its implementation and promotion in practical environment. This paper focuses on the above three problems, which is organized as follows: Section 2 represents the related research work on the above three key issues; in Section 3, the first issue is discussed, and the ontology-based modeling method is proposed to describe miner’s context information; in Section 4, the second issue is discussed, and the service configuration model is constructed to connect the context information with the information services; in Section 5, the third issue is discussed, and the computational experiment method is presented to verify whether the service system can play its due role in various accident scenarios; The last part gives the conclusion and future research priorities.

2. Background and motivation 2.1. The context-aware technology With the emergence of various sensors and the related perceptive software, the context-aware technology comes into being. The concept of the context-aware application is first proposed by Schilit et al. [5] in 1994. Its core is that the system can dig out more useful information about the served customers by means of sensors and then provide the most suitable services to them. By using the context-aware technology, the customer service experience can be improved a lot. Generally, context-aware service has the following four main features [6]:  Automation. Context-aware services are triggered by the data collected from sensors automatically, which can save the operators’ time and energy. It is especially suitable for miners working underground, which is dark, dangerous, and hard to predict.  24 h/7 d availability. Context-aware services are not restricted by manual working schedule. The served object can be monitored for 7  24 h.  Real-time response (Perceptiveness). The efficiency of contextaware services is only affected by the frequency of data acquisition, the speed of data transmission and the consumption time of service invocation. Generally, the instant service is very important to the emergency management in coal mine industry.

Model of context service Active context

Physical context

User context Resources of context

Passive context Fig. 1. The types of context-aware services.

In addition, through analyzing the historical data of the environment, the context-aware service can predict its trend and provide some proactive or preventive services for miners.  Personalized customization. Because the context-aware service is based on each served object’s context information, it is completely personalized. In addition, the data collection and the service strategy can also be configured according to the served object’s customized demands. As shown in Fig. 1, the context-aware services can be classified into four types based on context information sources as well as service models. According to Schmidt et al. [7], the context information can be categorized into User context and Physical context. Taking a miner as an example, the information about User context refers to his age, mood, habit, preference, ability, etc., while the information about Physical context includes time, location, temperature, light, facilities, etc. In accordance with the definition of Chen and Kotz [8], the service models can be categorized into active service mode (Active context) and passive mode (Passive context). When operating in the active service mode, the service system will change its behavior pattern in accordance with the received contextual information. For example, in a meeting, the phone will be switched into a silent mode according to the pre-set order. When operating in the passive service mode, the service system will change its service content in accordance with the received context information. For example: some goods can be recommended to a customer in accordance with his preferences when he is shopping online. At present, the context-aware technology has been used in many fields gradually, such as home environment, tourism, mobile computing, healthy care, etc. The Olivetti Active Badge Project [9] is the first application of the context-awareness technology, by means of which the automatic call forwarding service is realized in an office building. Bardram and Nørskov [10] presented a ContextAware Patient Safety and Information System (CAPSIS), which could monitor the surgical procedure in the operating room by using RFID tags on items and personnel. Wood et al. acquired the real-time data of the residents through integrating various sensors, such as environmental sensors, physiological sensors, and activity sensors etc. By learning their behavior patterns, the system can be used to assist the residential living [11]. Literature [12] and [13] emphasized the importance of users’ data and their context information in the recommendation system. Taking the mobile business model as an example, Reuver and Haaker [14] gave a detailed description on how to design a context-aware service and pointed out some problems in building a context-aware business model. Sridevi et al. [15] proposed a context-aware healthcare

X. Xue et al. / Computers in Industry 65 (2014) 291–305 Table 1 The abbreviation and full names of the information services in coal mine. Service

Type

Personal Positioning Service Gas Monitoring Service Oxygen Monitoring Service CO Monitoring Service Wind Speed Monitoring Service Temperature Monitoring Service Broadcasting Service Users’ Alarming Service Equipment Management Service Safe Center Alarming Service Safety hazards Troubleshooting Service Cooling Service Cooling Reminder Service Ventilation Reminder Service Emergency Rescue Services

Context information service PPS Context information service GasMS Context information service OxMS Context service COMS Context service WSMS Context service TmMS Business service BcS Business service UAS Business service EMS Business service SCAS Business service SHTS Business Business Business Business

Abbreviations

service service service service

CoS CRS VRS ERS

system architecture for connecting rural hospitals (with a high population density and a serious lack of appropriate medical facilities) with urban hospitals (well-equipped) in India, which can take advantage of the equipments and staff in urban hospitals to provide the remote medical services for the patient in rural hospitals according to his context information. 2.2. The concept of intelligent mines In recent years, the traditional mine industry in China has made a considerable progress. Currently, there has been a large amount of information systems in coal mine industry, which lay a solid foundation for the implementation of the context-aware services. As shown in Table 1, the existing information services in coal mine can be classified into two major categories: (1) the context information services, which are used to acquire environment data around the miner, e.g. miner’s location (the personal positioning service), the concentration of Gas (Gas Monitoring Service), wind speed (Wind Speed Monitoring Service) and so on; (2) the specialized business services, which are used to provide some key information to assist miner’s work, e.g. the hidden risks (Safety hazards Troubleshooting Service), the nearest locations of shelters (Emergency Rescue Services) and so on. However, the existing information systems are applied in different departments of coal mine industry, and their technical routes and deployed platforms vary widely. As a result, it is very hard to make full use of most information resources. On the one hand, information resources are abundant; on the other hand, the appropriate information services provided for miners are absent. Based on the above background, some researchers in China put forward the concept of ‘‘intelligent mines’’ [3], which borrowed the idea from ‘‘Smart Earth’’ [2]. Its goal is to make the entire coal mine more thoroughly instrumented, more comprehensively interconnected, and more intelligent than digital mines by means of applying numerous advanced technologies (such as service computing, cloud computing, internet of things, etc.). For the management staff, the intelligent mine can provide more effective decision support. For a majority of underground miners, it can provide information services to handle all kinds of hidden risks effectively and ensure the safety of miners to a great extent. The overall architecture of ‘‘intelligent mines’’ is divided into four layers: the perception layer, the communication layer, the platform layer and the application layer [3]. The perception layer is responsible for collecting information, mainly composed of various types of RFID tags, readers, and transducers. The communication layer is responsible for transmitting data from the perception layer to the application layer by means of network infrastructure. The platform layer is responsible for integrating different data

293

resources to form an overall monitoring platform. The application layer is responsible for invoking particular business services to meet the miner’s individual needs. In the architecture, the implementation of the context-aware service is shown in Fig. 2, specifically divided into three steps [4]:  The acquisition of context information. The access of miners’ context information relies on the invocation of various monitoring systems. First, the PPS (Personnel Positioning Service) is invoked to obtain the served miner’s location; based on miner’s location, the service system can invoke other monitoring services to access the direct context data, including gases concentrations, temperature and wind speed, etc. The frequency of data collection depends on the importance of context data. For example, the acquisition frequency of the dangerous gases concentration should be higher than others because they might threaten miner’s safety; the acquisition frequency of the underground temperature may be lower than others, because it changes little and does not threaten miner’s safety directly.  The reasoning of context information. After obtaining all the necessary direct context data, the complete context model of the served miner can be established through reasoning. Before conducting the reasoning, all information resources about miner’s context need to be abstracted, including the geographical environment of coal mine, various sensors, various types of equipments, all kinds of accidents etc. Furthermore, based on the relations between them, the detailed reasoning rules can be created to support the accurate invocation of information services.  The provision of information services. After identifying the miner’s status through reasoning, the system will query the service configuration model and determine which information services need to be invocated for current miner. For example, if the miner’s context information is identified as ‘‘O2 concentration, danger’’ by the service system, it will inquire the service configuration model and then choose ‘‘User’s Alarming Service’’ as an appropriate service for him. As a result, the alarming message is sent to the portable terminal device carried by the miner to arouse his attention. With the number of information systems increasing continuously in coal mine, more and more business services can be integrated with service system to enhance its ability to meet miners’ needs.

2.3. Research question and methodology In the intelligent mine, the served miners is the service requestor, who demands the timely, accurate and personalized information services to help him avoid various hidden risks and deal with the incidents correctly; the existing information systems can be used as service providers, which provide the information services through encapsulating the information resources as web services. Through establishing the link between context information and business services, the context-aware service system can be used as a bridge between the served miners and the service providers. So, how to realize the context-aware service system has become the key technology in building the intelligent mines. However, coal mine is a high-risk environment with a huge amount of data, in which the served objects need various specialized services to deal with the complex geological condition, the sensitive dangerous sources, and the serious accident effect, etc. Moreover, the relationship between various specialized services is complicated and dynamic. Aiming at some specific situations, different served objects may require different specialized service or service composition. The above reasons have made

X. Xue et al. / Computers in Industry 65 (2014) 291–305

294

Fig. 2. The workflow of Intelligent mine.

it difficult to implement the context-aware services in coal mine, which can be summarized as the following three questions. (1) Q1: How to depict the miner’s context? The problem is about the served object. Since the majority of the served objects are the underground miners, the accurate description of miners’ context is the important prerequisite of system operation. If their context can’t be depicted correctly, their situation or requirements can’t be identified precisely. As a result, it is impossible to provide the appropriate services for them. Therefore, how to depict the miner’s context exactly is a key problem to be solved. (2) Q2: How to provide customized services? The problem is about the service. Generally, there are a large number of miners working underground at the same time. They are located in different areas and responsible for accomplishing different tasks, which leads to the demands for different services. The served miner couldn’t benefit from the inappropriate or non-optimal services. The delayed services or the wrong services may cause panic among the miners and even threaten their lives. Therefore, how to provide the customized service to meet miner’s individual demands is another problem to be addressed. (3) Q3: How to test the validity of services? The problem is about the matching between the served objects and the service. The ultimate target of the intelligent service system is to guarantee the safety of underground miners. Therefore, the service system need to be tested repeatedly in different conditions (e.g. different geographical environment, different types of accidents, different available services etc.) to ensure its validity and availability. However, given the diversity and variability of coal mine accidents, with the addition of the ethical, social, economical and legal

implications, it is impossible to test whether the system can react correctly to all kinds of extreme conditions (e.g. gas explosion, gas outburst and other disasters) in real environment. As a result, how to verify the validity and availability of service system under different scenarios has become a key issue to be resolved. Currently, the research in this field is still in its initial stage, which lacks the in-depth research on the implementation of the service system. In order to promote the application of contextaware services in the coal mine, this paper will adopt the following methods to solve the above questions: (1) Q1: Currently, the ontology is a very popular method to describe the context, which can depict the situation easily and appropriately. Researchers have proposed some common context models and frameworks, which are used to solve some common questions, such as semantic representation, context-reasoning, context-classification and so on [7–9,16]. All the work has laid a solid foundation for the context modeling of coal mine. Based on the specific characteristics and actual demands of coal mine, the Coal Mine Semantic Sensor Network Ontology (CMSSN Ontology) is proposed to depict the context information of miners, which relies on the inheritance and reuse of SSN Ontology [17] (Semantic Sensor Network ontology) and DOLCE Ultra Lite (DUL). CMSSN Ontology can be suitable for a variety of situations in coal mine industry. (2) Q2: This paper establishes a loosely coupled relationship between context information and specialized services by defining a configuration model to realize the customized service invocation. (3) Q3: In this paper, a computational experiment method is put forward to solve this problem. By means of constructing a 3D virtual coal mine environment, the process of various coal mine

X. Xue et al. / Computers in Industry 65 (2014) 291–305

295

Table 2 The classes and properties in the CMSSN Ontology. Ontology

Classes

Properties

Derived from DUL

Object, PhysicalObject, PhysicalPlace, SocialObject, Concept, Parameter, UnitofMeasurement, Situation, Event, Process, InformationEntity, InformationObject, Abstract, Region, Amount, Quality Platform, System, Device, Sensor, SensingDevice, Observation, DeploymentRealatedProcess, Deployment, ObservationValue, QuantityObservationValue, Property, MeasurementProperty, Accuracy, Frequency, MeasurementRange Headentry, Tailentry, DiggingFace, Chamber, MiningFace, Observation_Gas, Observation_CO, Observation_Temp, Observation_WS, Observation_Oxy, GasOutput, COOutput, TempOutput, WSOutput, OxyOutput, ObservationValue_Gas, ObservationValue_CO, ObservationValue_Temp, ObservationValue_WS, ObservationValue_Oxy, State, PhysicalEntity

hasLocation, isLocationOf, isClassifiedBy, hasDataValue

Derived from SSN Ontology

Specialized for CMSSN Ontology

disasters can be simulated accurately and repeatedly. Thus, the performance of the service system can be analyzed and assessed in different experiments to determine whether the service system can meet miners’ actual needs.

3. The representation and reasoning of the context information The miner’s context information is mainly classified into the direct one and the indirect one [18]. The direct context information is obtained from all kinds of monitoring services (such as the Personal positioning system, Gas monitoring system and so on) in coal mine directly, including the location of miners, the concentration of various gases, temperature and humility, wind speed and pressure, the distribution of rescue capsules, etc. The indirect context information is the one obtained from the reasoning of miner’s direct context information, e.g. miner’s status is normal or abnormal. For example, based on miner’s current location and the concentration of various gases at the location, the miner’s safety situation can be determined by reasoning. Currently, it is a popular method to describe context information by means of ontology modeling, which can depict the context information easily and aptly. DOLCE (a Descriptive Ontology for Linguistic and Cognitive Engineering) is a standardized upper ontology, which is mainly used to describe the common objects in the real world [19]. DOLCE Ultra Lite (DUL) is the abridged version of DOLCE [20]. By ignoring the complex factors such as the time index, relation network and so on, it tries to lower the difficulty of modeling information concept ontology. SSN ontology (Semantic Sensor Network ontology) is developed by the W3C Semantic Sensor Network Incubator Group in 2009 [16]. In 2011, the final version was formed [17]. It provides a good framework to describe the context information by reusing the DUL ontology. However, SSN Ontology doesn’t provide the description of the domain concept, time, location, etc., but the description of sensors, sensor observations and related concepts [21]. Therefore, the application in some specific areas needs to be further expanded based on the upper ontology. By means of the inheritance of SSN Ontology and DOLCE Ultra Lite, combined with the domain characteristics and actual demand of coal mine, the Coal Mines Semantic Sensor Network Ontology (CMSSN Ontology) is built in this paper. As shown in Table 2, the CMSSN Ontology inherited some classes as well as some properties from the DUL and SSN Ontology. Meanwhile, according to the characteristics of the coal mine domain, 22 specialized classes and some attributes are constructed in the CMSSN to meet the basic needs of service system.

deployedSystem, deployedOnPlatform, onPlatform, hasDeployment, attachedSystem, isDeployed, hasSubSystem, isSubSystemOf, hasValue, observes, observedProperty, ObservationResult isPlatformOf, hasObservation, hasState, hasUnit, hasMaxValue, hasMinValue, classfiedAmount, hasAccuracy, hasFrequency, hasMeasurementRange, hasQuantityValue, isProducedBy

The class diagram and the inheritance relationship of CMSSN Ontology are shown in Fig. 3. On one hand, some classes in CMSSN are inherited directly from DUL and SSN, including the ‘‘PhysicalPlace’’ class in DUL, the ‘‘Property’’ class in SSN, the ‘‘SensingDevice’’ class in SSN and others. For example, in order to classify the location in coal mine, the ‘‘PhysicalPlace’’ class in DUL is inherited by CMSSN Ontology, which is further divided into several subclasses, including Transport tunnel, Return air tunnel, Chamber, Digging Face and Mining Face. If necessary, the subclasses can be further divided. On the other hand, the derived classes are analyzed and reused in CMSSN Ontology. For example, the ‘‘System’’ class in SSN Ontology is described as ‘‘parts of a sensing infrastructure, which has components, its subsystems, which are other systems.’’ [17]. According to the definition, all kinds of sensors in the same area (e.g. a Temperature sensor and a Gas sensor located in the same section of a tunnel) can be taken as a system and various sensors in the section can be considered as the subsystems. The entire CMSSN Ontology can be extended at any time according to the actual needs. In order to classify the major accidents dangerous sources and essential factors in coal mines, the ‘‘Property’’ class in SSN Ontology is inherited and redefined as ‘‘PhysicalEntity’’ class in CMSSN Ontology. The class can be instantiated to describe some common context information sources, such as Gas, CO, Oxy, Temp, WS, etc. In order to depict all kinds of sensors, the ‘‘SensingDevice’’ class in SSN Ontology is inherited by CMSSN Ontology. In the paper, all sensor instances in the working face 20711 and 20712 are derived from the class. The entire CMSSN Ontology can be extended at any time according to the actual needs. The relationships between Ontology concepts are represented by the property. The Ontology properties are divided into Object properties (ObjectProperty) and Data properties (DataProperty). Object properties and Data properties have their corresponding domains and ranges. Both the domain and the range of Object Property are defined as classes. Taking the Object property ‘‘hasLocation’’ as an example, its domain is the Platform class and its range is the ‘‘PhysicalPlace’’ class. The domain of Data Property is class type, while its range is the data type, such as integer, float, char and others. Taking the Data property ‘‘HasDataValue’’ as an example, its domain is the ‘‘ObservationValue’’ class and the ‘‘Amount’’ class (including the observed objects and a variety of observed properties, such as gas concentration, wind speed, temperature, and their accuracy, frequency, range, etc.), while its range is various data types (for example, the data type of gas concentration is float type). By means of the application of Data Property, the accuracy of data processing can be controlled easily in some certain cases, which can avoid some potential errors

296

X. Xue et al. / Computers in Industry 65 (2014) 291–305

Fig. 3. The class diagrams and inheritance relationship in CMSSN Ontology.

in the conversion of data types and ensure that the important data can be processed correctly. Besides some classes, the relationships between classes in the SSN Ontology are also inherited by CMSSN Ontology, including ‘‘deployedSystem’’, ‘‘deployedOnPlatform’’, ‘‘onPlatform’’, ‘‘hasDeployment’’, ‘‘attachedSystem’’, ‘‘isDeployed’’ and so on. Furthermore, some new relationships are created in the CMSSN Ontology according to the actual demands in coal mine industry. In Fig. 4, a very important reasoning process based on the relationships in CMSSN Ontology is marked with red line: PhysicalPlace ! Deployment ! Platform ! System ! Sensor. The detailed process is shown as follows: the sensor with ID ‘‘Sensor_20711_S1_Gas’’ is a part of the system with ID ‘‘System_20711_1’’; by means of a deployment process with ID ‘‘Deployment_20711_S1’’ the process, the system is deployed in a platform with ID ‘‘Platform_20711_1’’; according to the ‘‘IsLocationof’’ relation between the ‘‘Platform’’ and the ‘‘PhysicalPlace’’, the location of the platform is identified as ‘‘Headentry_20711_S1’’; in the end, the relationship between ‘‘Sensor’’ and ‘‘PhysicalPlace’’ is established. This reasoning process is critical for the implementation of service system, which can identify miner’s location based on the Personal Positioning Service (PPS), and then obtain the related indirect context value at the location based on the above reasoning process. This paper adopts OWL to describe miner’s context information, which is a kind of Web Ontology language proposed by W3C Web Ontology Work Group [22]. The context information is described as the RDF triples in the format of (subject, predicate, object). In the triples, the subject is the Ontology class; the object can be class, data or data type; and the predicate is the property defined in the Ontology. For example, the context information model (Headentry_20712_S2, isLocationOf, John) means that John is located in Headentry_20712_S2. The complex context model can be realized by composing several simple context models. For example, (John, LocatedIn, 20712workingFace) ^ (O2, HasValue, 21) means that John is located in 20712 working face where the O2 concentration

is 21%. (Sensor_20712_S3_Gas_ObservationValue, hasDataValue, 0.07) ^ (Sensor_20712_S3_Gas_ObservationValue, isClassifiedBy, %) mean that the observation value of Sensor_20712_S3_Gas is 0.07%. The OWL is strong at semantic expression and reasoning [23]. In the paper, the reasoning module is implemented with Jena which is developed by HP Labs [24]. The reasoning module includes OWL reasoning (the reasoning based on Ontology context model) and User-defined reasoning (the reasoning based on User-defined rules).  The OWL-based reasoning mainly relies on the relations between classes (e.g. disjoint, etc.), the cardinality (e.g. existence, arbitrary, minimum, maximum, completeness, etc.), the attributive character (e.g. symmetry, etc.), and the property type (e.g. numeration, etc.). The attributive character of rules should be predefined, such as the transferability and reversibility. Taking the transferability as an example, the property ‘‘subClassOf’’ can be adopted to define the inheritance between a generic class and a specific class. For example, A is a subclass of B, and B is a subclass of C, then A is a subclass of C. The property can be used to supplement some details of context information. For example, the miner ‘‘John’’ is located in the central water pump house, which belongs to a kind of chamber. After reasoning, it can be concluded that John is located in the chamber. The reasoning process is shown as follows: (John, LocatedIn, central water pump house), (central water pump house, subClassOf, chamber) ! (John, LocatedIn, chamber).  The User-defined reasoning adopts the forward-chaining rule engine. Taking a served miner as an example, the reasoning process of identifying his safety state is described as follows: (1) based on the miner’s current location, the reasoning process ‘‘PhysicalPlace ! Platform ! Deployment ! System ! Sensor (Sensor1, Sensor2 . . .)’’ is used to find out which sensors are at the location; (2) the reasoning process ‘‘Sensor ! Observation ! SensorOutput ! QuantityObservationValue ! DataProperty (hasDataValue)’’ is

X. Xue et al. / Computers in Industry 65 (2014) 291–305

297

Fig. 4. The relationship between some instances of CMSSN Ontology.

used to obtain the observed value of the sensors. Furthermore, based on the related specialized knowledge on coal mine industry, we can judge whether the observed values are normal or not. The detailed reasoning process is shown in Table 3. 4. The invocation of the customized service composition As is well known to all, the geological environment in coal mine is extremely complicated, which is sensitive to the changes of the context factors. As a result, any occasional small accident

might lead to other serious accidents and the huge loss. What’s more, the number of miners working underground at the same time is large. They are responsible for accomplishing different tasks, which leads to the demands for different services. There are many kinds of information services in intelligent coal mine (as shown in Table 1). In order to meet the specific needs of different users, how to provide the customized service to meet miner’s individual demands is a key problem to be addressed. Furthermore, it is important to emphasize the timeliness of the customized service invocation. The delayed service or the wrong

Table 3 Part of reasoning rules and reasoning instances. Reasoning rules

Reasoning instances

[Rule_Loc: (?a, rdf:type, PhysicalPlace) (?a, isLocationOf,?b) (?b, isPlatformOf,?c) (? c, deployedSystem,?e) (?e, hasSubsystem,?f) ! (?a, isLocationOf,?f)] [Rule_Value: (?a, hasObservation,?b) (?b, ObservationResult,?c) (?c, hasValue,?d) (?d, hasDataValue,?e) ! (?a, hasObservationValue,?e)]

a = Headentry_20712_S2, b = Platform_20712_S2, c = Deployment_20712_S2, d =System_20712_S2, e =Sensor_20712_S2_Gas ! (Headentry_20712_S2, isLocationOf, Sensor_20712_S2_Gas) a = Sensor_20712_S2_Gas, b = Observation_20712_S2_Gas, c =Sensor_20712_ S2_GasOutput, d = Sensor_20712_S2_GasOutput_value, e = 1.1 ! (Sensor_20712_S2_Gas, hasObservationValue, 1.1) a = Headentry_20712_S2, b = Sensor_20712_S2_Gas, c = Sensor_20712_S2_ GasOutput_value, d = 1.1 ! (Sensor_20712_S2_Gas, hasState, abnormal)

[Rule_State: (?a, isLocationOf,?b) (?a, rdf:type, Headentry) (?b, rdf:type, Sensor) (?b, hasObservationValue,?c) (?c, hasDataValue,?d) lessthan (?d, 1.0) ! (?b, hasState, abnormal)]

298

X. Xue et al. / Computers in Industry 65 (2014) 291–305

Fig. 5. The context-aware-based customized service invocation in intelligent coal mine.

service may cause panic among the miners and even threaten their lives. The context-aware-based customized service selection and invocation is shown in Fig. 5. After analyzing and classifying the received context information, the appropriate business service or service composition can be invoked in accordance with its type and priority. Fig. 6 gives an example on how to invoke the appropriate service to deal with the abnormal gas concentration. When receiving the gas concentration data at some monitoring point, it will be compared with the risk thresholds: if the gas concentration

data is slightly overweight but still in a safe range (a < X  b), the Ventilation Notification Service (VNS) will be called to notify the operators of ventilation system without informing miners working at the location; when the gas concentration data exceeds the alarm value but still in the controllable range (b < X  c), the Ventilation Notification Service (VNS) will be invoked to reduce gas concentration, and the Users’ Alarming Service (UAS) is invoked to warn miners at the same time; when the gas concentration data exceeds the alarm value (c < X), the Ventilation Notification Service (VNS), the Users’ Alarming Service (UAS) and the

Fig. 6. The example of the customized service invocation.

X. Xue et al. / Computers in Industry 65 (2014) 291–305

299

Table 4 The classification of context information and the corresponding service configuration model. Type

The division of context data

CH4

In the main return air tunnel

In the working face

In the transport tunnel

CO

O2

Temp

Everywhere The value is 10  20 ppm The value is 20  80 ppm The value is >80 ppm The value is 20.0021.00% The value is 19.0020.00% The value is <19.00% In working face

In electronical chamber

Wind Speed

In the transport tunnel, working face, intake air tunnel, return air tunnel

In the main intake air tunnel

In other ventilating tunnels

The The The The The The The The The The

value value value value value value value value value value

is is is is is is is is is is

0.50  0.70% 0.70  1.00% >1.00% 0.70  1.00% 1.00  1.50% >1.50% >0.30  0.50% 0.50  1.00% >1.00% 5.0016 .00%

The value is >25.00 8C The value is 25.00  30.00 8C The value is 330.00 8C The value is 30.00  33.00 8C The value is 33.00  36.00 8C The value is 336.00 8C The abnormal value is 5.006.00 m/s or 0.250.35 m/s The abnormal value is >6.00 m/s or <0.25 m/s The abnormal value is 7.008.00 m/s The alerting value is >8.00 m/s The value is <0.15 m/s

Degree

Service composition

Abnormal, but within the safe range Beyond the alert, but under control Dangerous, and beyond the control Abnormal, but within the safe range Beyond the alert, but under control Dangerous, and beyond the control Abnormal, but within the safe range Beyond the alert, but under control Dangerous, and beyond the control Extremely dangerous, explosion Abnormal, but within the safe range Beyond the alert, but under control Dangerous, and beyond the control Abnormal, but within the safe range Beyond the alert, but under control Dangerous, and beyond the control Abnormal, but within the safe range Beyond the alert, but under control Dangerous, and beyond the control Abnormal, but within the safe range Beyond the alert, but under control Dangerous, and beyond the control Abnormal, but within the safe range

SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! ERS ! SHTS ! EMS ! VNS SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! ERS ! SHTS ! EMS ! VNS SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! ERS ! SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! ERS ! SHTS ! EMS ! VNS SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! ERS ! SHTS ! EMS ! VNS SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! ERS ! SHTS ! EMS ! VNS SHTS ! EMS ! CNS (UAS, BCS, SCAS) ! SHTS ! EMS ! CNS (UAS, BCS, SCAS) ! ERS ! SHTS ! EMS ! CNS SHTS ! EMS ! CNS (UAS, BCS, SCAS) ! SHTS ! EMS ! CNS (UAS, BCS, SCAS) ! ERS ! SHTS ! EMS ! CNS SHTS ! EMS ! VNS

Beyond the alert, but under control

(UAS, BCS, SCAS) ! SHTS ! EMS ! VNS

Beyond the alert, but under control

(UAS, BCS, SCAS) ! SHTS ! EMS ! VNS

Dangerous, and beyond the control

(UAS, BCS, SCAS) ! ERS ! SHTS ! EMS ! VNS

Beyond the alert, but under control

(UAS, BCS, SCAS) ! SHTS ! EMS ! VNS

Note: The service in the brackets are invoked in parallel, ! indicates the order of calling service.

Emergency Rescue Service (ERS) will be called to help miners to leave the location safely and timely. The connection between context information and business service is the key to realize the customized service invocation. If the invocation is based on the Ontology reasoning completely, it will be limited by the performance of the reasoning engine. It is difficult for service system to deal with a large amount of context data and business services in time. In the paper, the reasoning rules were transformed into the service configuration model (shown in Table 4). The context information and the business services are defined respectively in the table, which is more flexible and can support the reuse of context information. In this way, the reasoning is substituted by querying the service configuration model in the database, which can reduce the costs and save the time a lot. In this configuration model, the value of context data is designed strictly in accordance with the Safety Regulations in Coal Mine. In order to facilitate the application of the rules, the continuous context data should be discretized (e.g. the concentration of coal dust, temperature, etc.). The above configuration model is only an incomplete version, which can become perfect by adding the new dangerous sources (e.g. the underground fire, flooding, gas outburst, tunnel collapse etc.) continuously according to the actual demands in the future. It is important to emphasize that the connection between context information and business services is not just a simple one-to-one relationship. There are two special cases which need to be paid attention to: (1) A change in context data may trigger multiple service invocations. Since miners differ in their jobs, there are many kinds of miners such as equipment inspection and maintenance staff, mining auxiliary staff, safety inspectors, leaders of a working

field and so on. These miners have their specific duties or tasks and need the correspondent information services. Therefore, one kind of context information may involve more than one type of miners, who need different types of services. Taking the most commonly used ‘‘alarm’’ service as an example, when the gas concentration at the miner’s location is more than the alert value, the served objects include the miner, his partners at the location, and the safety supervision center. So, the three related services (UAS, BCS and SCAS) need to be invoked at the same time. (2) There exists the priority order in the process of service invocation. Because there is a sequential relationship between the services, the service invocation needs to obey a certain priority order in accordance with their operation regulations in coal mine industry. Assuming that ‘‘the failure of ventilation equipment causes the abnormal gas concentration in the tunnel’’, the invoked service composition is shown as follows: (1) Invoke the ‘‘Safety hazards Troubleshooting Service’’ to notify the inspectors to identify the causes of gas concentration exceeding; (2) Invoke the ‘‘equipment maintenance service’’ to notify the maintenance personnel of the possible hidden risk in the equipments; (3) Invoke ‘‘ventilation notification service’’ to notify the device operator of increasing the wind speed to reduce the gas concentration in the tunnel. Because the miner moves randomly, it is very difficult to predict when and how the context data will change. Therefore, the ‘‘polling–calling’’ mechanism is used to trigger the invocation of business services. When the context data is beyond a certain threshold, the business services will be invoked. The workflow of the service system is shown in Fig. 7, and the detailed steps are shown as follows:

300

X. Xue et al. / Computers in Industry 65 (2014) 291–305

Fig. 7. The workflow of service system.

Step1 Initialization. Initialized parameters: Location – the miner’s location; Type – the miner’s job type; Cxt – the miner’s context data; CV – the miner’s state; SS – the service set; KC – the key context data with the highest priority; SC – the service composition (orchestration or choreography). Step2 The acquirement of context information. AcquireSystemState (system_state); If (system_state == Error) /* judge the state of system */ { SCAS (system(state), Error); /* Alarm the Supervision Center that the system goes wrong*/ Sleep(2000); /* sleep for 2 second */ Goto AcquireSystemState; } Input: Miner_ID/*acquire miner’s ID from the coal mine working records.*/

AcquireLocation (Miner_ID)/*invoke PPS to get the miner’s location */ Output: Location Input: Location AcquireContextInfo (Location); /*Select and invoke context services to acquire context data */ Output: Cxt Step3 The reasoning of context information. Input: Cxt AnalyzeContextInfo (Cxt); /* classifying the context data */ Output: CV; /*miner’s state*/ If (CV.value =0) /*no danger*/ { Sleep(2000); Goto Step2; /*after a fixed interval, check the context again*/ } Else { Goto Step4; /* Invoke business services*/ }

X. Xue et al. / Computers in Industry 65 (2014) 291–305

Step4 The selection of the key context. Input: CV; For (i = 0; i < K; i ++) /*K is the number of the miner’s context data*/ { If (CV i .time > the limit) /* the out-of-date context data is deleted from the queue*/ Delete the CV i from the queue; KC = MaxPriority(CV i ); /*KC is the context data with the highest priority*/ } Output: KC; Step5 The selection of service set. Input: KC; SS = QueryConfiguration(KC); /*Query Configuration Model for the appropriate service*/ Output: SS Step6 The Execution of service composition. Input: SS If(SS.length ! = 0) { SC = CompositeService(SS) /* form service composition based on the priority algorithm.*/ Execute(SC); } If (queue.number ! = 0) /*the queue is not empty*/ Goto Step 4;

Step7 Goto Step2

301

geographical environment, characteristics of the disaster, miners’ distribution, etc) on the validity of service system. As a result, computational experiment can be adopted as a new means to provide a comprehensive, accurate, quantitative analysis and evaluation of service system for its further improvement. As shown in Fig. 8, based on the geographic data and monitoring data of a state-owned coal mine (May 2011–June 2011), a 3D coal mine simulation environment is constructed, including the environment module (e.g. the network structure of mine tunnel, the current location of excavation, the direction of ventilation, etc.), the context module (e.g. the real-time concentration of gas, CO and other harmful gases humidity, wind speed, etc.), the miner module (e.g. the number, composition, and distribution of miners in each underground area), the equipment module (e.g. main fan, local fan, damper, shearer, belts, pumps, vehicles, air compressors, etc.) and so on. A variety of disaster scenarios can be simulated on the simulation platform, e.g. gas outburst, gas explosions, flooding accident, and so on. Then, based on the specialized agent model, the miner agents will produce a series of emergency response and escape behavior to deal with the disaster in the virtual environment. At the end of each execution cycle, the status of the agents and the environment will be updated and shown in the 3D virtual coal mine environment. Thus, the performance of service system in emergency treatments can be displayed intuitively. In the intelligent coal mine, the alarming service is one of the most important context-aware services. In the case study, we will test whether the service system can react to the abnormal Gas concentration correctly. The following is the detailed steps: (1) The setting of computational experiment

5. The verification of service system based on computing experimentation There are so many complicated safety regulations in coal mine, which requires very high accuracy and reliability in service invocation. However, due to the diversity and the variability of coal mine accidents, with the addition of the impact of the economic, legal, and ethical, it is impossible to test whether the system can react correctly to all kinds of extreme conditions (e.g. gas explosion, gas outburst and other disasters) in real environment. Thus, how to verify whether the served miners get the appropriate services has become an obstacle to the implementation of service system. Based on the above background, this paper proposes the method of computational experiment to solve this problem. By designing different ‘‘experiments’’ on the 3D virtual mine simulation platform, the process of various coal mine accidents can be reappeared accurately. Thus, the operation of service system under different situations can be analyzed and evaluated to determine whether its performance can be satisfactory. Compared with the existing experimental methods, the advantages of the computational experiment method are shown as follows: (1) Computational experiment is precisely controllable. By setting the parameters of the environment (e.g. all kinds of geographical factors, the distribution of miners, and so on) as well as the disaster (including time, location, type, scale and so on), the process of various coal mine accidents can be reappeared accurately and the operation of service system can be analyzed in a quantitative mode. (2) Computational experiment is easy to operate. In the simulation environment, it is easy to implement a variety of extreme experiments (e.g. gas explosion, gas outburst, fire and other disasters) to find out the potential problems in service system, such as the correctness of the invoked service, the efficiency of service selection, etc. (3) Computational experiment is repeatable. This facilitates designing different computational experiments to assess the influence of various factors (such as

In the configuration interface of miners (top left of Fig. 8), the miners’ parameters and the related services can be configured. The miner’s parameter contains his basic information (e.g. his ID, name, task, age, etc.) and his advanced information (e.g. his initial location, walking route, walking speed, etc.). By means of setting all these parameters, the goal and the actions of each miner agent can be customized and initialized. The related services of the miner agent include the context information services (e.g. personnel positioning service, Gas Monitoring Service etc.) and the specialized business services (e.g. Users’ Alarming Service etc.). In the configuration interface of reasoning rules (middle left of Fig. 8), the rules can be added, modified and deleted. In the case study, the rules are defined as: (1) If one sensor in the personnel positioning system obtains the miner’s ID, the miner’s location will be identified as the sensor’s location. (2) If the miner’s location is determined, various monitoring services can be invoked to obtain the other context data at the location, such as gas concentration, oxygen concentration, CO concentration, temperature, wind speed, and so on. (3) If the gas concentration around the miner is beyond a certain value, the alarming service will be invoked to send alarming messages to the miner. In the configuration interface of environment parameters (bottom left of Fig. 8), the environment parameters (e.g. O2 concentration, CH4 concentration, CO concentration, temperature and so on) in the experiment can be set, including the initial value, the range, the trend and so on. At the same time, the outbreak of the abnormal accident can be set, including its initial location, value, type, degree and so on. (2) The operation of computational experiment The overall operation of coal mine as well as various miners’ working scenarios can be exhibited in the 3D Virtual Coal Mine. In the scene of the whole coal mine (middle top of Fig. 8), by

302

X. Xue et al. / Computers in Industry 65 (2014) 291–305

Fig. 8. The configuration and the operation interface of the virtual coal mine simulation platform.

dragging the mouse or clicking the control buttons at the upper right corner, the macro scenery of the experiment can be observed from various perspectives. In the scene of the local tunnel (middle bottom of Fig. 8), the reaction of the miner agents to accident can be shown in an intuitive way.

Meanwhile, the miner’s current context data (top right of Fig. 8), the status of service system (middle right of Fig. 8) and the operation data of the invoked services (bottom right of Fig. 8) are displayed in the right-hand column of the system.

Fig. 9. The miner’s route.

X. Xue et al. / Computers in Industry 65 (2014) 291–305

303

Fig. 10. The context data chart along miner’s route.

After completing the configuration, the experiment starts to run. In Fig. 9, the miners’ walking route is shown as ‘‘A ! B ! C ! D ! E’’, i.e. from 20712 transport tunnel to 20712 working face. In Fig. 10, all kinds of context data along the miner’s route is shown. The related details and the invoked services at each point are shown in Table 5. It can be seen that the abnormal data (Gas-State, Danger) appears at the location B-t6, which needs to call the corresponding service composition. Through comparing the actual invoked services (in Table 5) and the desired service invocation (in Table 2), the operation of service system can be analyzed in detail. (3) The analysis of computational experiment

The validity of service invocation mainly refers to whether the service system can make the proper response to all abnormal accidents. The verification of validity can ensure the miners can be provided with the correct services in all abnormal situations. This directly determines the value of the service system. Here, v indicates the validity of the service system; R indicates the total number of service invocation; A indicates the number of the correct service invocation. Thus, the validity of service system can be expressed as:

In this paper, the assessment of the context-aware service system includes the validity and availability of service invocation. The availability of service invocation refers to whether the service system could response to all the abnormal context data timely. The verification of availability can ensure the miner can be provided with the timely services to deal with all abnormal situations. Here, a indicates the availability of the service system; E indicates the total number of the simulated abnormal situations; R indicates the total number of service invocation. Thus, the availability of service system can be expressed as:

According to Table 4, five different types of accidents are designed to test the validity and availability of service system, which occur 600 times randomly in the virtual environment. According to the statistics of the operation data, the availability of service systems can be calculated according to Formula (1). As shown in Fig. 11, the availability of service system in the four abnormal situations (i.e. the gas, CO, oxygen and wind speed) maintains an average rate of 80% with a slight fluctuation. However, because the priority of handling the temperature abnormal is lower than the above four situations, its process is often delayed when the other four kinds of abnormal situations occur at the same time. In some cases, if the delay is beyond a certain time, this temperature abnormal will be omitted without any process. As a result, the availability of service

a¼

R E

(1)

v¼

A R

(2)

Table 5 The acquired context data and the corresponding service composition. Node

O2 (%)

CO (ppm)

Gas (%)

T (8C)

WS (m/s)

A B-t1 B-t2 B-t3 B-t4 B-t5 B-t6 B-t7 B-t8 B-t9 C D E

20.9 20.6 20.5 20.5 20.3 20.1 20.0 19.8 20.4 20.5 20.4 20.5 20.5

3 5 5 4 4 5 4 4 5 6 7 6 6

0.00 0.37 0.18 0.28 0.30 0.59 0.70 1.10 0.88 0.69 0.51 0.57 0.34

23.23 24.21 24.23 24.24 24.22 24.25 24.25 24.27 24.27 24.31 24.29 23.30 24.27

1.9 1.8 1.7 1.7 1.8 1.7 1.6 1.7 1.6 1.7 1.7 1.8 1.7

State

(Gas-State, danger) (Gas-State, danger) (Gas-State, danger)

Business service composition No No No No No No SHTS ! EMS ! VNS (UAS, BCS, SCAS) ! SHTS ! EMS ! VNS SHTS ! EMS ! VNS No No No No

X. Xue et al. / Computers in Industry 65 (2014) 291–305

304

Fig. 11. The availability of Service System.

Fig. 12. The validity of service system.

system to the abnormal Temp situation is about 70%, which is lower than the average level. According to Table 4, if the invoked service is not the desired one or is not a complete service composition, it will be considered as an incorrect service. According to the statistics of operation data of the service system, the validity of service system can be calculated according to Formula (2). As shown in Fig. 12, the validity of service system in the five abnormal situations (i.e. the gas, CO, oxygen, wind speed and Temp) maintains an average rate of above 80% with a slight fluctuation. The validity of the system in a single abnormal situation maintains a high rate of 97%. However, the rate is low in two or more concurrent abnormal situations. The improvements on this problem will be focused in further study. 6. Conclusion With the rapid development of information technologies in China, more and more information systems have been deployed in the coal mine industry in recent decades. However, the served objects are mainly confined to the management staff. For a majority of underground miners, it is difficult for them to benefit from the informatization of coal mine. As a result, those hidden risks cannot be prevented effectively, which leads to a great waste of information resources. The context-aware-based intelligent service system is a suitable choice to solve the above problem, which can provide miners with the customized services based on their real-time situations. Currently, the research in this field is still in its initial stage, which lacks detailed discussions and in-depth researches on the

implementation of the service system, especially the following three key issues: (3) how to depict the miner’s context; (2) how to provide customers with customized services; (3) how to verify the validity of the service system. In this paper, the above three problems are discussed in detail: (1) For the first issue, based on the specific characteristics and actual demands of coal mine, the Coal Mine Semantic Sensor Network Ontology (CMSSN Ontology) is proposed to depict the context information of miners. The Ontology relies on the inheritance and reuse of SSN Ontology (Semantic Sensor Network ontology) and DOLCE Ultra Lite (DUL), which can be suitable for a variety of situations in coal mine industry. (2) For the second issue, this paper defines a configuration model to establish the connection between context information and specialized services, which facilitates realizing the customized service invocation. (3) For the third issue, the computational experiment method is put forward to test the validity and availability of the service system. By means of constructing a 3D virtual coal mine environment, the process of various coal mine disasters can be simulated accurately and repeatedly. Thus, the performance of the service system can be analyzed and assessed in a comprehensive way, which can help us to determine whether the service system can meet miners’ actual needs. This paper is an exploration on how to apply the context-aware technology in coal mine industry, which provides some key technical support for the construction of ‘‘intelligent coal mine’’. In the future, we will further strengthen the research in each module of service system, including how to detect and verify the false alarms through multi-sensor data fusion; how to analyze and predict the trend of context data and realize the more accurate

X. Xue et al. / Computers in Industry 65 (2014) 291–305

early warning; how to provide the special information services to deal with the extreme abnormal situations (e.g. underground fire, flooding, gas outburst, tunnel collapse etc.) and so on. Acknowledgements The work is supported by the National Natural Science Foundation of China (No. 61175066, No. 60905041, No. 51174263, No. 61300124); Program for Science & Technology Innovation Talents in Universities of Henan Province (No. 2012HASTIT013); Foundation for The Excellent Youth Scholars of University of Henan Province (No. 2011GGJS-056); Program for Innovative Research Team of Henan Polytechnic University (T2013-3); Key Laboratory of Mine Spatial Information Technologies of SBSM (KLM201110). References [1] State administration of coal mine safety, China coal industry yearbook 2011, China coal industry publishing house, Publication Date: July 2012. [2] S.J. Palmisano, A Smarter Planet. The Next Leadership Agenda: Address to the Council on Foreign Relations (6th November), Council on Foreign Relations, New York, 2008. [3] Y.Z. Sun, B. Wang, Research on the construction and application of intelligent mine, 2012, http://www.cb.com.cn/1634427/20120702/391576.html. [4] X. Xue, J.K. Chang, The Research on Context-Aware-Based Intelligent Service System for Miners, 2012 IEEE Ninth International Conference on Services Computing (SCC2012), (2012), pp. 478–485. [5] B. Schilit, N. Adams, R. Want, Context-aware computing applications, Workshop on Mobile Computing Systems and Applications(WMCSA), (1994 (December)), pp. 85–90. [6] T. Mo, W.P. Li, Z.H. Wu, W.J. Chu, Framework of context-awareness based service system, Chinese Journal of Computers 33 (11) (2010 (November)) 2083–2093. [7] A. Schmidt, M. Beigl, H.W. Gellersen, There is more to context than location, Computers & Graphics 23 (6) (1999 (December)) 893–901.

305

[8] G. Chen, D. Kotz, A survey of context-aware mobile computing research, 2000 (November) (Tech. Rep. TR2000-381, Dartmouth). [9] R. Want, A. Hopper, V. Falco, et al., The active badge location system, ACM Transactions on Information Systems 10 (1) (1992) 10–91. [10] J.E. Bardram, N. Nørskov, A context-aware patient safety system for the operating room, Proceedings of the 10th international conference on Ubiquitous computing, (2008), pp. 272–281. [11] A. Wood, et al., Context-aware wireless sensor networks for assisted living and residential monitoring, IEEE Network 22 (4) (2008) 26–33. [12] S. Abbar, M. Bouzeghoub, S. Lopez, Context-aware recommendation systems: a service-oriented approach, 35th International Conference on Very Large Data Bases (VLDB), Lyon, France, (2009), pp. 1–6. [13] G. Adomavicius, A. Tuzhilin, Context-Aware recommender systems, Recommender Systems Handbook Part 1 (2011) 217–253. [14] M. De Reuver, T. Haaker, Designing viable business models for context-aware mobile services, Telematics and Informatics 26 (2009) 240–248. [15] S. Sridevi, B. Sayantani, K. Amutha, C. Madan Mohan, R. Pitchiah, Context Aware health monitoring system, Lecture Notes in Computer Science 6165 (2010) 249– 257. [16] W3C Semantic Sensor Network Incubator Group, Major products of the group’s work, July 2011, 2011, http://www.w3.org/2005/Incubator/ssn/wiki/Main_Page. [17] P. Barnaghi, M. Compton, O. Corcho, et al., Semantic Sensor Network XG Final Report [R], W3C Incubator Group Report 28 (2011) http://www.w3.org/2005/ Incubator/ssn/XGR-ssn-20110628/. [18] T. Gu, H.K. Pung, D.Q. Zhang, A Service-Oriented middleware for building ContextAwareness services, Elsevier Journal of Network and Computer Applications 28 (1) (2005) 1–18. [19] C. Masolo, S. Borgo, A. Gangemi, et al. The WonderWeb Library of Foundational Ontologies Preliminary Report [R], WonderWeb Deliverable (D17), 2003. [20] The Laboratory for Applied Ontology (LOA), DOLCE-2.1-Lite-Plus (2013) http:// www.loa.istc.cnr.it/DOLCE.html. [21] W3C Semantic Sensor Network Incubator Group, Semantic Sensor Network Ontology (2013) http://www.w3.org/2005/Incubator/ssn/ssnx/ssn. [22] W3C: Web Ontology Language, http://www.w3.org/TR/2009/REC-owl2-overview-20091027/, 2009. [23] J. Moskal, C. Matheus, Detection of suspicious activity using different rule engines – Comparison of BaseVISor, Jena and Jess Rule Engines. RuleML ‘08 Proceedings of the International Symposium on Rule Representation, Interchange and Reasoning on the Web 5321 (2008) 73–80. [24] D. Reynolds, Jena 2 Inference support (2010) http://openjena.org/inference/ index.html.