- Email: [email protected]
An Ontology-based Model for Typical-context Awareness in the Oil Products Distribution System
Available online at www.sciencedirect.com
ScienceDirect Procedia Computer Science 96 (2016) 1156 – 1165
20th International Conference on Knowledge Based and Intelligent Information and Engineering Systems, KES2016, 5-7 September 2016, York, United Kingdom
An Ontology-based Model for Typical-context Awareness in the Oil Products Distribution System D
D
/LMXQ6XQ )DQJIDQJ/L ;LDQJSHL+X a
D
Faculty of Management and Economics, Dalian University of Technology, No.2 Linggong Road Gaoxinyuan District, Dalian City 116023, China
Abstract
During the oil products distribution process, disruptions are diverse and can occur at different distribution states and the distribution state changes constantly, which makes it hard to capture the distribution context in real time. In this paper, an ontology-based model is developed, which consists of the disruption ontology, the distribution-state ontology and the SWRL rules. The model can capture the typical-context in real time and judge whether the original distribution plan is disrupted by the disruption under the current context, which is in favor of taking measures to respond to the disruption in real time. © 2016 2016The TheAuthors. Authors. Published Elsevier © Published by by Elsevier B.V.B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of KES International. Peer-review under responsibility of KES International Keywords: Typical-context;
Ontology; SWRL
1. Introduction During the oil products distribution process, disruptions (such as gas stations changing their demands, disabled roads induced by traffic accidents or traffic jams, vehicle breakdowns) are diverse and can occur at different distribution states. If we can’t readjust vehicle routes in real time to respond to the disruptions, the distribution process will be interrupted. We called the instability state of the distribution system caused by disruptions a typicalcontext, which may cause interruption and other serious consequences, since oil products, as an important production and living materials, must be distributed in time. Otherwise, the social production and citizen’s living would fall in chaos. Thus, it is important to capture typical-context in real time to handle disruptions timely and finally guarantee the delivery process smoothly.
* Corresponding author. E-mail
address: [email protected]
1877-0509 © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of KES International doi:10.1016/j.procs.2016.08.158
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
1157
In the oil products distribution system, the core of the typical-context awareness is to capture the distribution context in real time and judge the disruption to the system caused by the disruptions. It’s hard to do so, as the distribution state changes constantly with the distribution process. Moreover, disruptions are diverse and can occur at different distribution states. In practice, experienced schedulers can respond to different kinds of disruptions promptly. However, they often use intuitive approach to make a simple adjustment, which may not feasible or often be accompanied by inaccuracies and unsuitable for large-scale problems. In order to achieve the typical-context awareness, this paper proposes an ontology-based model, which represents the distribution context using ontology and judge whether the original distribution plan is disrupted by disruptions under the current context using the SWRL rules. As the ontology can represent the objects involved in the distribution system and the complex relationship between the objects, it can support capturing the disruptions and the distribution state in real time. Meanwhile, the ontology has the advantage of reasoning, which is beneficial to realize the integration of knowledge representation and reasoning. Thus, the proposed ontology-based model is applicable to judge whether the original distribution plan is disrupted by the disruption under the current context. The remaining part of the paper is organized as follows. Section 2 reviews the related work. Section 3 proposes the framework of the ontology-based model. Finally, in section 4, conclusions are drawn and further study is introduced. 2. Related work In the oil products distribution system, typical-context awareness is to judge whether the original distribution plan is disrupted by the regular disruption under the current context, which aims to respond to a disruption in real time with a satisfied solution. The solution is not necessarily with the lowest cost but it should try to avoid bringing more disturbances to the current system. Hence it is related to the research area of the disruption management on the distribution, which is reviewed in section 2.1. Besides that, the building of the ontology-based model is related to the research area of ontology-based knowledge representation method, which is reviewed in section 2.2. 2.1 Disruption management on the distribution Existing researches with regard to disruption management in the area of distribution focus on the following aspects: algorithms and models. Representative study abroad on model and algorithm is as follows. With regard to the problem of real-time customer requests and dynamic travel time, Potvin et al. [1] proposed a solution whose objective is to minimize the sum of the vehicle travel time, the delay time of serving the customer and the delay time of returning to the central warehouse. With respect to the problem of vehicle breakdowns, Li et al. [2, 3, 4] developed a Lagrangian relaxation based-heuristic to obtain the strategy for rescheduling one or more vehicles to serve a particular trip and other service trips originally scheduled for the disabled vehicle. With respect to the problem of new customer requests, Ichoua et al. [5] proposed a vehicle diversion strategy and employed the Tabu search algorithm to solve it. Domestic scholars’ study on model and algorithm is as follows. For the vehicle routing problem with time Windows, Mingchun Wang [6] put forward a disruption recovery strategy. For the four types of disruptions in the process of logistics distribution, Jiewei Song and GangRong [7] built the integer programming model and well dealt with the impact of the uncertain factors using the genetic algorithm with heuristic algorithm. Based on the perspective of behavior, Qiulei Ding [8] developed a logistics distribution disruption management model with the changed customer time-window. Considering both customers and logistics managers, Xiangpei Hu et al. [9] established a disruption model to measure the delay of time. Xuping Wang et al. [10], Xiangpei Hu et al. [11] studied distribution disruption management problems with the fuzzy time window and the changed demand respectively. With respect to the problem of travel time delaying, Zheng Wang et al. [12] developed the disruption management model and algorithm on distribution vehicle scheduling. At present, most of these modeling methods use a specific model and algorithm to solve a specific kind of disruption under a predefined distribution state. And they can only solve static problems and can’t handle the diversified unexpected events in real time. Xiangpei Hu [13] and Lijun Sun[14] put forward an intelligent modeling method based on knowledge representation to solve the distribution disruption management problem in real time, which provided a new means of computer modeling to automatically handle the disruption. However, oil products
1158
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
distribution is with high safety requirements, huge delivery quantity and other characteristics, which needs further research on knowledge-based intelligent modeling method. 2.2 The ontology-based knowledge representation methods This research is to build an ontology-based model for typical-context awareness in the oil products distribution system, which is related to the ontology in the field of context awareness. Besides, the complex reasoning knowledge is needed to be represented based on the ontology-based model. Therefore, it involves two research areas, the ontology in the field of context awareness and the reasoning methods based on ontology. 2.2.1 The ontology in the field of context awareness According to the context content described by the ontology, Juan Ye et al. [15] thought that the ontology can be divided into the ontologies of time, location, person/agent, resources, trajectory and activity. As for the time ontology, the W3C OWL-Time ontology provides a vocabulary for Allen’s [16]thirteen temporal relations and some other common time ontology are DAML - Time [17], OpenCyc [18], RCC [19], and theReiPolicy [20] Ontology. As for the location ontology, the ASC’s [21] location ontology is mainly used for modeling spatial positions in these coordinate systems and for representing distances between positions. In contrast, the primary goal of the CONON location ontology is to differentiate between indoor and outdoor spaces. As for the actor ontology, Kabiretal [22] developed the Social Context Information Management System (SCIMS), which is based on a core upper ontology that defines four first-class entities: Person, SocialRole, Relationship, and CurrentStatus, and domain specific ontologies can be extended from this four aspects. Besides, FOAF is an ontology centered on linking people not only through social relationships, but also through networks of human collaboration and interests. As for the resource ontology, the GoodRelations ontology [23] provides an e-commerce vocabulary for describing products and services. As for the trajectory ontology, ontologies have been applied to try to compress and abstract raw GPS data into higher-level qualitative information. One stream of the research is to represent a trajectory into a sequence of stops and moves [24, 25]. The above five types of ontology is to describe a single object. However, an event is defined as a generic concept and links with temporal, spatial, and actor ontologies via corresponding properties. Existing domain-independent ontologies for event modeling, like ontology SOUPA [26], Ontonym and The Event Ontology (EO), are presenting the different design patterns and expressive capabilities. In summary, the ontology in the field of context awareness can represent the concepts and the relations between concepts, which is beneficial to describing the domain knowledge. Event ontology can also be used to describe the high-level, complex activities in reality. However, in the oil products distribution system, as unexpected events are diversified and the distribution state changes constantly with the plan executing process, while oil products distribution system involves more context elements and the relationship between objects is more complex, more applicable ontology-based model based on these characteristics needs to be developed. 2.2.2 The reasoning methods based-on ontology Since McDermott and Allen put forward the time reasoning in Artificial Intelligence (AI) [27, 28], the ontology reasoning has been extended from the field of AI to the fields of semantic, knowledge representation, semantic similarity and so on [29, 30, 31]. At present, there are mainly two kinds of ontology reasoning mechanism: the reasoning mechanism based on description logic ontology and the reasoning mechanism based on rules [32]. (1)The reasoning mechanism based on description logic ontology The research on ontology reasoning mechanism based on description logic mainly concentrates in the improvement of the algorithm aiming to improve the efficiency of ontology reasoning. The Commonly used ontology reasoning algorithm is the Structural Subsumption algorithm [33] and Tableau algorithm [34, 35]. The general method to test the logical satisfaction is the Tableau algorithm. As Tableau algorithm has good decidability and completeness, it is widely used in the reasoning based on description logic. Hence, many researchers make improvement to Tableau algorithm [35, 36, 37, 38]. The ontology reasoning based on description logic is very suitable for representing the field of application by concept taxonomy. However, because it can’t provide rules for reasoning, it is difficult to express the relationship
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
1159
between attributes of classes. The reasoning provided by description logic is limited [39]. (2)The reasoning mechanism based on rules SWRL language is the common language in the ontology reasoning mechanism based on rules. At present, the related research is mainly on the improvement to the undecidable characteristic of SWRL, such as, the limitation of grammar rules [40], as well as the cross between rules and ontology. Ontology reasoning based on rules is suitable for the ontology which has stronger rules and has more expression on rules. It is relatively simple. If it is applied to large-scale ontologies, it couldn’t guarantee the high efficiency of reasoning. This research is to represent the complex reasoning knowledge. By selecting the rule-based reasoning model, we can describe the suitable inference rules by SWRL language. Then, we can combine the SWRL rules and the knowledge base of ontology to achieve the reasoning. However, as the ontology-based model is large in the oil products distribution system, the reasoning efficiency of the rule-based reasoning model is low. Further research is needed. 3. The ontology-based model of typical-context awareness In the oil products distribution system, the typical-context is decided by the characteristic of an unexpected event and the current distribution state. For example, when the gas station increases its demand suddenly, we need to consider three cases respectively: (1) If the gas station has been distributed, the demand-increasing disruption will result in a disruption to the current distribution plan, which needs schedulers to readjust vehicle routes in real time; (2) If the gas station is going to be distributed or is being distributed, and the increased demand is less than the redundant load of the vehicle which undertakes the gas station’s distribution tasks, it will not result in a disruption to the current distribution plan; (3) However, if the increased demand exceeds the redundant load of the vehicle, it will result in a disruption to the current distribution plan. A new distribution plan should be generated by scheduling other vehicles in real time. Thus, in order to capture the typical-context, we first need to capture the unexpected event and the current distribution state in real time, and then judge whether the original distribution plan is disrupted by the event under the current context. This paper achieves the typical-context awareness by developing the ontology-based model. Firstly, we build the unexpected-event ontology to represent the diversified unexpected events that ever happened in the oil products distribution system. Then, we build the distribution-state ontology to represent the constantly changed system state. And finally, we build the SWRL rules to represent the experienced reasoning knowledge. The typical context-aware process based on ontology model (cf. Fig. 1) is as follows: Step 1: when an unexpected event occurs, the ontology of the unexpected event and the distribution state is matched in the ontology base according to the characteristics of the unexpected event; Step 2: According to the characteristic of the unexpected event and the current state in the ontology, we can match the rules in the SWRL rules base; Step 3: According to the selected rules, we can judge whether the original distribution plan is disrupted by the unexpected event under the current context. To realize the above mentioned process, section 3.1 builds the unexpected-event ontology, while section 3.2 builds the distribution-state ontology, and section 3.3 builds SWRL rules.
1160
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
Fig. 1 An ontology-based typical-context awareness frame
3.1. The unexpected-event ontology In oil products distribution system, a distribution plan in execution is composed of three types of objects: gas stations, roads and vehicles. Thus, unexpected events can only come from the three objects. In order to classify the diversified unexpected events, we should first clarify each object can generate what kinds of unexpected events. As for the object "gas station", in the distribution process, a station may change its demand, including demand reducing or increasing. Besides, a station may also change the delivery time, including postponing delivery time and advancing delivery time. The unexpected events from the object "road" include disabled road and recoverable road caused by traffic accidents, traffic congestion, road construction and other Force Majeure. The unexpected events from the object "vehicle" are vehicle breakdowns, including unrecoverable breakdowns and recoverable breakdowns at a short time. From the above, in order to represent unexpected events, firstly we should determine the corresponding relationships between unexpected events and objects. Secondly, according to the impact to the objects from the unexpected event, we can determine the subclass of the event. And finally, we need a data field to record the parameter information related to it. For example, for a demand-increasing unexpected event, the increased demand quantity and the gas station's ID, of which brought out the event should be recorded, and for a time-delaying event, the delayed time of the road and the ID of the road, or the vehicle ID of the vehicle which occurred a breakdown should be recorded. The data field is called "value". Above all, we use the object attributions of the ontology to represent the corresponding relationship between unexpected events and objects and the inclusion relation between the unexpected event and its subclasses. Besides, we use the data attribute of the ontology to record the unexpected events’ parameter information. For example, Fig.2 describes the road-sourcing unexpected-event ontology.
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
1161
Fig.2 The road-sourcing unexpected-event ontology
Specifically, in the ontology, "hasRoadEvent" represents the road that sources the unexpected event. "is-a" represents the unexpected event contains two subclasses: Disabled Road and Recoverable Road, "has" represents the Disabled Road has the Road ID and the Recoverable Road has the Road ID and Delayed Time. 3.2. The distribution-state ontology A distribution plan is composed of three types of objects: gas stations, roads and vehicles. So the distribution state can be reflected by the three objects. Although the state of an object can be reflected by many aspects, we just concerned those closely related to the distribution process and the states of objects can be represented as the date attributions of the ontology (Fig.3). For the gas station, when an unexpected event occurs, the gas station may have been served or not been served or being served. We can use "served" "unserved" or "being served" respectively to express the three states. For the road, when an unexpected event occurs, the road may have been traveled by the planned vehicle, or have not been traveled, or be being traveled. We use the "traveled" "untraveled" or "being traveled" to express the three states. For the vehicle, when an unexpected event occurs, the vehicle may be executing a distribution plan or be lying idle. We use the "busy" and "idle" to express the two states. The state of the object can be obtained by the corresponding relationship between objects and the relative positions of two types of objects in the distribution system. For example, in a route, when a vehicle is between a gas station and the last served gas station, the state of the current gas station is "unserved", and the state of the last served gas station is “served”, and the state of the road between the two gas stations is "being traveled". When a vehicle is located in a gas station, the state of the gas station is "being served" and the state of the road between this station and the last served station is "traveled". When a vehicle returns to the distribution center after serving the last station, the state of the vehicle will be from "busy" to "idle". When a vehicle travels from the distribution center to the first station, the state of the vehicle will be from "idle" to "busy". Therefore, in the ontology, in order to capture the state of the object, we need to represent the corresponding relationship between the three objects: gas stations, roads and vehicle. Besides, we need to represent the location information of each object. The objects-relationship ontology is shown in Fig. 3.
1162
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
Fig. 3 The objects-relationship ontology
Specifically, "serve" represents that the relationship between the object "vehicle" and the object "GasStation", which means that the gas station will be served by the vehicle and the vehicle serves all the stations with the same vehicle ID. "belong" represents the relationship between the object "road" and the object "route" , which means that a road is part of a route and the roads which have the same route ID belong to the same route. "travel" represents the relationship between the object "vehicle" and the object "route" , which is a pair relationship. It means that a vehicle can only travels one specific route. "link" represents the relationship between the object "road" and the object "gas station", which means that a road links two gas stations. The object "road" refers to the shortest path between two stations, and the object "route" is a collection of "road". "GSLocation" "RLocation" and "Vlocation" represent the positions of gas stations, roads and vehicles respectively. 3.3. SWRL Rules The knowledge to judge whether the original distribution plan is disrupted by an unexpected event under the current context comes from the experienced scheduler. Such knowledge can be represented as a series of rules in the form of "IF Condition THEN Action". Thus, we use SWRL rules which combine rules and ontology to represent such knowledge. By SWRL, we can use the vocabularies and the relationships depicted in ontology to describe rules. Such rules are called Disruption Analysis and Judgment Rules. According to the characteristics of the unexpected event we can divide the Disruption Analysis and Judgment Rules into six types. For example, we call the rules which are used to analyze the change of demand as "Demand-Change-Judging Rule", which is a rule set containing six rules shown in Table 1. The other five types of rules are "Time-Window-Change-Judging Rule", "Disabled-Road-Judging Rule", "Road-Time-Delay-Judging Rule", "Disabled-Vehicle-Judging Rule", "ReparableVehicle-Judging Rule".
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
Table 1 "Demand-Change-Judging Rule" Rule
Rule Content
Specific Meaning
Rule1
GasStationState (? x ) ġ hasState (? x, served ) ġ UnexpectedEvent(?y) ġ hasEvent (? y, DemandChange)ĺinfluence (? y, disturbsystem)
If the gas station changes the demand at the time when the gas station’s state is "served", the unexpected event will cause a disruption to the distribution system.
Rule2 GasStationState (? x ) ġ hasState (? x, being-served ) ġ UnexpectedEvent(?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: moreThan (?z, 0) ġ swrlb: moreThan (?z, rl) ĺinfluence (? y, disturbsystem) GasStationState (? x ) ġ hasState (? x, unserved ) ġ UnexpectedEvent(?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: moreThan (?z, 0) ġ swrlb: moreThan (?z, rl) ĺinfluence (? y, disturbsystem)GasStationState (? x ) ġ hasState (? x, unserved ) ġ UnexpectedEvent(?y) ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: moreThan (?z, 0) ġswrlb: moreThan (?z, rl) ĺinfluence (? y, disturbsystem) Rule3 GasStationState (? x ) ġ hasState (? x, being-served) ġ UnexpectedEvent(?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: moreThan (?z, 0) ġ swrlb: lessThan (?z, rl) ĺinfluence (? y, nodisturbsystem) GasStationState (? x ) ġ hasState (? x, unserved) ġ UnexpectedEvent(?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: moreThan (?z, 0) ġ swrlb: lessThan (?z, rl) ĺinfluence (? y, nodisturbsystem)
Rule4 GasStationState (? x ) ġ hasState (? x, being- served ) ġ UnexpectedEvent (?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: lessThan (?z, 0) ġ swrlb: sameAs (?z, d) ėinfluence (? y, disturbsystem) GasStationState (? x ) ġ hasState (? x, unserved ) ġ UnexpectedEvent (?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: lessThan (?z, 0) ġ swrlb: sameAs (?z, d) ėinfluence (? y, disturbsystem) Rule5 GasStationState (? x ) ġ hasState (? x, being- served ) ġ UnexpectedEvent (?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: lessThan (?z, 0) ġ swrlb: lessThan (?z, d) ėinfluence (? y, nodisturbsystem) GasStationState (? x ) ġ hasState (? x, unserved ) ġ UnexpectedEvent (?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: lessThan (?z, 0) ġ swrlb: lessThan (?z, d) ėinfluence (? y, nodisturbsystem)
If the gas station increases the demand and the increased demand exceeds the redundant load of the vehicle at the time when the gas station’s state is "unserved" or "being served", the unexpected event will cause a disruption to the distribution system.
If the gas station increases the demand and the increased demand is less than the redundant load of the vehicle at the time when the gas station’s state is "unserved" or "being served", the unexpected event will not cause a disruption to the distribution system. If the gas station reduces the demand and the reduced demand equals to the quantity of the order at the time when the gas station’s state is "unserved" or "being served", the unexpected event will cause a disruption to the distribution system. If the gas station reduces the demand and the reduced demand is less than the quantity of the order at the time when the gas station’s state is "unserved" or "being served", the unexpected event will not cause a disruption to the distribution system.
1163
1164
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
4. Conclusion During the oil products distribution process, unexpected events are diverse and can occur at different distribution states and the distribution state changes constantly, which make it hard to capture the distribution context in real time. In this paper, an ontology-based model is developed, which build the disruption ontology, the distribution-state ontology and the SWRL rules to capture the typical-context in real time. The ontology-based model can capture the typical-context in real time and judge whether the original distribution plan is disrupted by the unexpected event under the current context, which is in favor of taking measures to respond to the disruption in real time. The model in this paper could just judge whether the original distribution plan is disrupted by an unexpected event under the current context. The policies responding to the disruptions in real time has not been considered, which is important for recovering the distribution. Therefore, further research should focus on the building of the policies. Acknowledgements This work has been partially supported by the grants from Natural Science Foundation of China (Nos. 71201014, 71571027, 71272093, 71271037); The Fundamental Research Funds for the Central Universities (Nos. DUT14QY28ˈDUT14RC(4)04)
References 1. Potvin JY, Xu Y, Benyahia I. Vehicle routing and scheduling with dynamic travel times. Computers & Operations Research 2006;4:11291137. 2. Li JQ, Mirchandani PB, Borenstein D. Real-time vehicle rerouting problems with time windows. European Journal of Operational Research 2009;3:711-727. 3. Li JQ, Mirchandani PB, Borenstein D. A Lagrangian heuristic for the real-time vehicle rescheduling problem. Transportation Research Part E 2009;3:419-433. 4. Li JQ, Borenstein D, Mirchandani PB. A decision support system for the single-depot vehicle rescheduling problem. Computers & Operations Research 2007;4:1008-1032 5. Ichoua S, Gendreau M, Potvin JY. Diversion issues in real-time vehicle dispatching. Transportation Science 2000;4:426-438. 6. Wang MC, Gao CX, Zeng YT. Recovery of the VRPTW disruption and the TABU SEARCH algorithm. Journal of Mathematics 2006;2:231236. (in Chinese) 7. Song JW, Rong G. Study of uncertainty problem in vehicles scheduling. Journal of Zhejiang University (Engineering Science) 2003;2:243248.(in Chinese) 8. Ding QL. A model of disruption management with the changes of time windows-based on the perspective of behavior. Chinese Journal of Management Science 2015;5:89-97. (in Chinese) 9. Hu XP, Yu N, Ding QL. Sequential decision methods for disruption management in distribution. Journal of Industrial Engineering /Engineering Management 2011;2:186-190, 130. (in Chinese) 10. Wang XP, Ruan JH, Zhang K, et al. Study on combinational disruption management for vehicle routing problem with fuzzy time windows. Journal of Management Sciences in China 2011;6:2-15. (in Chinese) 11. Hu XP, Sun LJ, Wang YN. A model for disruption management in urban distribution systems. Journal of Management Sciences in China 2011;1:50-60. (in Chinese) 12. Wang Z, Hu XP, Wang XP. Disruption management model and algorithm for distribution vehicle scheduling problems under accidental travel time delay. Systems Engineering-Theory & Practice 2013;2:378-387. (in Chinese) 13. Hu XP, Sun LJ, Liu LL. A PAM approach to handling disruptions in real-time vehicle routing problems. Decision Support Systems 2013;3:1380-1393. 14. Sun LJ, Hu XP. A Knowledge Representation Method for Algorithms in DSS for Real-time Vehicle Routing in Urban Distribution. International Journal of Innovative Computing, Information and Control 2012;8:5859-5872. 15. Ye J, et al. Semantic web technologies in pervasive computing: A survey and research roadmap. Pervasive and Mobile Computing 2015;23:125. 16. Allen JF. Towards a general theory of action and time. Artificial Intelligence 1984;2:123-154. 17. Hobbs JR, Pan F. An ontology of time for the semantic web. ACM Trans. Asian Lang. Inf. Process 2004;1:66-85. 18. Lenat DB, Guha RV. Cycling paper reviews-response. Artificial Intelligence 1993;1:149-174. 19. Borgo S, Guarino N, Masolo C. A pointless theory of space based on strong connection and congruence. In: Aiello LC, Doyle J, Shapiro S, editors. Proceedings of the fifth international conference. San Mateo: Morgan kaufmann; 1996. p. 220-229.
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
20. Kagal L, Finin T, Joshi A. A policy based approach to security for the semantic web. In: Fensel D, Sycara K, Mylopoulos J, editors. Proceedings of the second International Semantic Web Conference. Berlin: Springer; 2003. p. 402-418. 21. Strang T, Linnhoff-Popien C, Frank K. Applications of a context ontology language. In: Begusic D, Rozic N, editors. Proceedings of the International Conference on Software. Croatia: Univ. Split; 2003. p. 14-18. 22. Kabir M, Han J, Yu J, Colman A. Scims: a social context information management system for socially-aware applications. In: Ralyte J, Brinkkemper S, Wryczr S, editors. Proceedings of 24th International Conference. Berlin: Springer; 2012. p. 301-317. 23. Hepp M. Goodrelations: An ontology for describing products and services offers on the web. In: Gangemi A, Euzenat J, editors. Proceedings of the 16th International Conference on Knowledge Engineering: Practice and Patterns. Berlin: Springer; 2008. p. 329-346. 24. Yan Z, Chakraborty D, Parent C, Spaccapietra S, Aberer K. Semantic trajectories: mobility data computation and annotation. ACM Trans. Intell. Syst. Technol 2013;3:1-38. 25. Spaccapietra S, Parent C, Damiani ML, DeMacedo JA, Porto F, Vangenot C. A conceptual view on trajectories. Data Knowl. Eng. 2008;1:126-146. 26. Chen H, Perich F, Finin T, Joshi A. SOUPA: standard ontology for ubiquitous and pervasive applications. In: ieee computer society, editors. Proceedings of Mobiquitous 2004. Los Alamitos: IEEE Computer; 2004. p. 258-267. 27. McDermott D. A temporal logic for reasoning about process and plans. Cognitive Science 1982;2:101-155. 28. Allen JF. Towards a general theory of action and time. Artificial Intelligence 1984;2:123-154. 29. Qu YZ. A predicate-ordered logic for knowledge representation on the web. Future Generation Computer Systems 2004;1:19-26. 30. Riccardo R. On the decidability and complexity of integrating ontologies and rules. Web Semantics: Science, Services and Agents on the World Wide Web 2005;1:61-73. 31. Lin X, Liu WY. Context based knowledge representation and reasoning-from the perspective of artificial intelligence. Computer Science 2005;1:142-146. (in Chinese) 32. Li HY, Wang YB, Qu YZ, Pan JZ. A Reasoning Algorithm for pD*. In: Riichiro M, Zhong ZS, Fausto G, editors. Proceedings of First Asian Semantic Web Conference. Berlin: Springer; 2006. p. 293-299. 33. Wen KM. Research on Semantic Search Based on Ontology Repository Reasoning. Wuhan: Huazhong University of Science and Technology, 2007. (in Chinese) 34. Badder F, Sattler U. Tableau Algorithms for Description Logics. In: Dyckhoff R, editors. Lecture Notes in Artificial Intelligence. Berlin: Springer; 2000. p. 1-18. 35. Horrocks I, Sattler U. A Tableaux decision procedure for SHOIQ. Journal of automated reasoning 2007;3:249-276. 36. Peng S. Research on ontology reasoning based on description logic in semantic web. JiLin: Jilin University, 2009. (in Chinese) 37. Badder F, Sattler U. An overview of tableau algorithms for description logics. Studia Logica 2001;1:5-40. 38. Horrocks I, Patel-Schneider PF. Evaluating optimized decision procedures for propositional modal Kappa((m)) satisfiability. Journal of automated reasoning 2002;2:173-204. 39. ter Horst HJ. Extending the RDFS Entailment Lemma. In: McIlraith SA, Plexousakis D, VanHarmelen F, editors. Proceedings of 3rd Int. Semantic Web Conference. Berlin: Springer; 2004. p. 77-91. 40. Motik B, Sattler U, Studer R. Query answering for OWL-DL with rules. J. Web Semantic 2005;1:41-60.
1165
ScienceDirect Procedia Computer Science 96 (2016) 1156 – 1165
20th International Conference on Knowledge Based and Intelligent Information and Engineering Systems, KES2016, 5-7 September 2016, York, United Kingdom
An Ontology-based Model for Typical-context Awareness in the Oil Products Distribution System D
D
/LMXQ6XQ )DQJIDQJ/L ;LDQJSHL+X a
D
Faculty of Management and Economics, Dalian University of Technology, No.2 Linggong Road Gaoxinyuan District, Dalian City 116023, China
Abstract
During the oil products distribution process, disruptions are diverse and can occur at different distribution states and the distribution state changes constantly, which makes it hard to capture the distribution context in real time. In this paper, an ontology-based model is developed, which consists of the disruption ontology, the distribution-state ontology and the SWRL rules. The model can capture the typical-context in real time and judge whether the original distribution plan is disrupted by the disruption under the current context, which is in favor of taking measures to respond to the disruption in real time. © 2016 2016The TheAuthors. Authors. Published Elsevier © Published by by Elsevier B.V.B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of KES International. Peer-review under responsibility of KES International Keywords: Typical-context;
Ontology; SWRL
1. Introduction During the oil products distribution process, disruptions (such as gas stations changing their demands, disabled roads induced by traffic accidents or traffic jams, vehicle breakdowns) are diverse and can occur at different distribution states. If we can’t readjust vehicle routes in real time to respond to the disruptions, the distribution process will be interrupted. We called the instability state of the distribution system caused by disruptions a typicalcontext, which may cause interruption and other serious consequences, since oil products, as an important production and living materials, must be distributed in time. Otherwise, the social production and citizen’s living would fall in chaos. Thus, it is important to capture typical-context in real time to handle disruptions timely and finally guarantee the delivery process smoothly.
* Corresponding author. E-mail
address: [email protected]
1877-0509 © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of KES International doi:10.1016/j.procs.2016.08.158
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
1157
In the oil products distribution system, the core of the typical-context awareness is to capture the distribution context in real time and judge the disruption to the system caused by the disruptions. It’s hard to do so, as the distribution state changes constantly with the distribution process. Moreover, disruptions are diverse and can occur at different distribution states. In practice, experienced schedulers can respond to different kinds of disruptions promptly. However, they often use intuitive approach to make a simple adjustment, which may not feasible or often be accompanied by inaccuracies and unsuitable for large-scale problems. In order to achieve the typical-context awareness, this paper proposes an ontology-based model, which represents the distribution context using ontology and judge whether the original distribution plan is disrupted by disruptions under the current context using the SWRL rules. As the ontology can represent the objects involved in the distribution system and the complex relationship between the objects, it can support capturing the disruptions and the distribution state in real time. Meanwhile, the ontology has the advantage of reasoning, which is beneficial to realize the integration of knowledge representation and reasoning. Thus, the proposed ontology-based model is applicable to judge whether the original distribution plan is disrupted by the disruption under the current context. The remaining part of the paper is organized as follows. Section 2 reviews the related work. Section 3 proposes the framework of the ontology-based model. Finally, in section 4, conclusions are drawn and further study is introduced. 2. Related work In the oil products distribution system, typical-context awareness is to judge whether the original distribution plan is disrupted by the regular disruption under the current context, which aims to respond to a disruption in real time with a satisfied solution. The solution is not necessarily with the lowest cost but it should try to avoid bringing more disturbances to the current system. Hence it is related to the research area of the disruption management on the distribution, which is reviewed in section 2.1. Besides that, the building of the ontology-based model is related to the research area of ontology-based knowledge representation method, which is reviewed in section 2.2. 2.1 Disruption management on the distribution Existing researches with regard to disruption management in the area of distribution focus on the following aspects: algorithms and models. Representative study abroad on model and algorithm is as follows. With regard to the problem of real-time customer requests and dynamic travel time, Potvin et al. [1] proposed a solution whose objective is to minimize the sum of the vehicle travel time, the delay time of serving the customer and the delay time of returning to the central warehouse. With respect to the problem of vehicle breakdowns, Li et al. [2, 3, 4] developed a Lagrangian relaxation based-heuristic to obtain the strategy for rescheduling one or more vehicles to serve a particular trip and other service trips originally scheduled for the disabled vehicle. With respect to the problem of new customer requests, Ichoua et al. [5] proposed a vehicle diversion strategy and employed the Tabu search algorithm to solve it. Domestic scholars’ study on model and algorithm is as follows. For the vehicle routing problem with time Windows, Mingchun Wang [6] put forward a disruption recovery strategy. For the four types of disruptions in the process of logistics distribution, Jiewei Song and GangRong [7] built the integer programming model and well dealt with the impact of the uncertain factors using the genetic algorithm with heuristic algorithm. Based on the perspective of behavior, Qiulei Ding [8] developed a logistics distribution disruption management model with the changed customer time-window. Considering both customers and logistics managers, Xiangpei Hu et al. [9] established a disruption model to measure the delay of time. Xuping Wang et al. [10], Xiangpei Hu et al. [11] studied distribution disruption management problems with the fuzzy time window and the changed demand respectively. With respect to the problem of travel time delaying, Zheng Wang et al. [12] developed the disruption management model and algorithm on distribution vehicle scheduling. At present, most of these modeling methods use a specific model and algorithm to solve a specific kind of disruption under a predefined distribution state. And they can only solve static problems and can’t handle the diversified unexpected events in real time. Xiangpei Hu [13] and Lijun Sun[14] put forward an intelligent modeling method based on knowledge representation to solve the distribution disruption management problem in real time, which provided a new means of computer modeling to automatically handle the disruption. However, oil products
1158
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
distribution is with high safety requirements, huge delivery quantity and other characteristics, which needs further research on knowledge-based intelligent modeling method. 2.2 The ontology-based knowledge representation methods This research is to build an ontology-based model for typical-context awareness in the oil products distribution system, which is related to the ontology in the field of context awareness. Besides, the complex reasoning knowledge is needed to be represented based on the ontology-based model. Therefore, it involves two research areas, the ontology in the field of context awareness and the reasoning methods based on ontology. 2.2.1 The ontology in the field of context awareness According to the context content described by the ontology, Juan Ye et al. [15] thought that the ontology can be divided into the ontologies of time, location, person/agent, resources, trajectory and activity. As for the time ontology, the W3C OWL-Time ontology provides a vocabulary for Allen’s [16]thirteen temporal relations and some other common time ontology are DAML - Time [17], OpenCyc [18], RCC [19], and theReiPolicy [20] Ontology. As for the location ontology, the ASC’s [21] location ontology is mainly used for modeling spatial positions in these coordinate systems and for representing distances between positions. In contrast, the primary goal of the CONON location ontology is to differentiate between indoor and outdoor spaces. As for the actor ontology, Kabiretal [22] developed the Social Context Information Management System (SCIMS), which is based on a core upper ontology that defines four first-class entities: Person, SocialRole, Relationship, and CurrentStatus, and domain specific ontologies can be extended from this four aspects. Besides, FOAF is an ontology centered on linking people not only through social relationships, but also through networks of human collaboration and interests. As for the resource ontology, the GoodRelations ontology [23] provides an e-commerce vocabulary for describing products and services. As for the trajectory ontology, ontologies have been applied to try to compress and abstract raw GPS data into higher-level qualitative information. One stream of the research is to represent a trajectory into a sequence of stops and moves [24, 25]. The above five types of ontology is to describe a single object. However, an event is defined as a generic concept and links with temporal, spatial, and actor ontologies via corresponding properties. Existing domain-independent ontologies for event modeling, like ontology SOUPA [26], Ontonym and The Event Ontology (EO), are presenting the different design patterns and expressive capabilities. In summary, the ontology in the field of context awareness can represent the concepts and the relations between concepts, which is beneficial to describing the domain knowledge. Event ontology can also be used to describe the high-level, complex activities in reality. However, in the oil products distribution system, as unexpected events are diversified and the distribution state changes constantly with the plan executing process, while oil products distribution system involves more context elements and the relationship between objects is more complex, more applicable ontology-based model based on these characteristics needs to be developed. 2.2.2 The reasoning methods based-on ontology Since McDermott and Allen put forward the time reasoning in Artificial Intelligence (AI) [27, 28], the ontology reasoning has been extended from the field of AI to the fields of semantic, knowledge representation, semantic similarity and so on [29, 30, 31]. At present, there are mainly two kinds of ontology reasoning mechanism: the reasoning mechanism based on description logic ontology and the reasoning mechanism based on rules [32]. (1)The reasoning mechanism based on description logic ontology The research on ontology reasoning mechanism based on description logic mainly concentrates in the improvement of the algorithm aiming to improve the efficiency of ontology reasoning. The Commonly used ontology reasoning algorithm is the Structural Subsumption algorithm [33] and Tableau algorithm [34, 35]. The general method to test the logical satisfaction is the Tableau algorithm. As Tableau algorithm has good decidability and completeness, it is widely used in the reasoning based on description logic. Hence, many researchers make improvement to Tableau algorithm [35, 36, 37, 38]. The ontology reasoning based on description logic is very suitable for representing the field of application by concept taxonomy. However, because it can’t provide rules for reasoning, it is difficult to express the relationship
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
1159
between attributes of classes. The reasoning provided by description logic is limited [39]. (2)The reasoning mechanism based on rules SWRL language is the common language in the ontology reasoning mechanism based on rules. At present, the related research is mainly on the improvement to the undecidable characteristic of SWRL, such as, the limitation of grammar rules [40], as well as the cross between rules and ontology. Ontology reasoning based on rules is suitable for the ontology which has stronger rules and has more expression on rules. It is relatively simple. If it is applied to large-scale ontologies, it couldn’t guarantee the high efficiency of reasoning. This research is to represent the complex reasoning knowledge. By selecting the rule-based reasoning model, we can describe the suitable inference rules by SWRL language. Then, we can combine the SWRL rules and the knowledge base of ontology to achieve the reasoning. However, as the ontology-based model is large in the oil products distribution system, the reasoning efficiency of the rule-based reasoning model is low. Further research is needed. 3. The ontology-based model of typical-context awareness In the oil products distribution system, the typical-context is decided by the characteristic of an unexpected event and the current distribution state. For example, when the gas station increases its demand suddenly, we need to consider three cases respectively: (1) If the gas station has been distributed, the demand-increasing disruption will result in a disruption to the current distribution plan, which needs schedulers to readjust vehicle routes in real time; (2) If the gas station is going to be distributed or is being distributed, and the increased demand is less than the redundant load of the vehicle which undertakes the gas station’s distribution tasks, it will not result in a disruption to the current distribution plan; (3) However, if the increased demand exceeds the redundant load of the vehicle, it will result in a disruption to the current distribution plan. A new distribution plan should be generated by scheduling other vehicles in real time. Thus, in order to capture the typical-context, we first need to capture the unexpected event and the current distribution state in real time, and then judge whether the original distribution plan is disrupted by the event under the current context. This paper achieves the typical-context awareness by developing the ontology-based model. Firstly, we build the unexpected-event ontology to represent the diversified unexpected events that ever happened in the oil products distribution system. Then, we build the distribution-state ontology to represent the constantly changed system state. And finally, we build the SWRL rules to represent the experienced reasoning knowledge. The typical context-aware process based on ontology model (cf. Fig. 1) is as follows: Step 1: when an unexpected event occurs, the ontology of the unexpected event and the distribution state is matched in the ontology base according to the characteristics of the unexpected event; Step 2: According to the characteristic of the unexpected event and the current state in the ontology, we can match the rules in the SWRL rules base; Step 3: According to the selected rules, we can judge whether the original distribution plan is disrupted by the unexpected event under the current context. To realize the above mentioned process, section 3.1 builds the unexpected-event ontology, while section 3.2 builds the distribution-state ontology, and section 3.3 builds SWRL rules.
1160
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
Fig. 1 An ontology-based typical-context awareness frame
3.1. The unexpected-event ontology In oil products distribution system, a distribution plan in execution is composed of three types of objects: gas stations, roads and vehicles. Thus, unexpected events can only come from the three objects. In order to classify the diversified unexpected events, we should first clarify each object can generate what kinds of unexpected events. As for the object "gas station", in the distribution process, a station may change its demand, including demand reducing or increasing. Besides, a station may also change the delivery time, including postponing delivery time and advancing delivery time. The unexpected events from the object "road" include disabled road and recoverable road caused by traffic accidents, traffic congestion, road construction and other Force Majeure. The unexpected events from the object "vehicle" are vehicle breakdowns, including unrecoverable breakdowns and recoverable breakdowns at a short time. From the above, in order to represent unexpected events, firstly we should determine the corresponding relationships between unexpected events and objects. Secondly, according to the impact to the objects from the unexpected event, we can determine the subclass of the event. And finally, we need a data field to record the parameter information related to it. For example, for a demand-increasing unexpected event, the increased demand quantity and the gas station's ID, of which brought out the event should be recorded, and for a time-delaying event, the delayed time of the road and the ID of the road, or the vehicle ID of the vehicle which occurred a breakdown should be recorded. The data field is called "value". Above all, we use the object attributions of the ontology to represent the corresponding relationship between unexpected events and objects and the inclusion relation between the unexpected event and its subclasses. Besides, we use the data attribute of the ontology to record the unexpected events’ parameter information. For example, Fig.2 describes the road-sourcing unexpected-event ontology.
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
1161
Fig.2 The road-sourcing unexpected-event ontology
Specifically, in the ontology, "hasRoadEvent" represents the road that sources the unexpected event. "is-a" represents the unexpected event contains two subclasses: Disabled Road and Recoverable Road, "has" represents the Disabled Road has the Road ID and the Recoverable Road has the Road ID and Delayed Time. 3.2. The distribution-state ontology A distribution plan is composed of three types of objects: gas stations, roads and vehicles. So the distribution state can be reflected by the three objects. Although the state of an object can be reflected by many aspects, we just concerned those closely related to the distribution process and the states of objects can be represented as the date attributions of the ontology (Fig.3). For the gas station, when an unexpected event occurs, the gas station may have been served or not been served or being served. We can use "served" "unserved" or "being served" respectively to express the three states. For the road, when an unexpected event occurs, the road may have been traveled by the planned vehicle, or have not been traveled, or be being traveled. We use the "traveled" "untraveled" or "being traveled" to express the three states. For the vehicle, when an unexpected event occurs, the vehicle may be executing a distribution plan or be lying idle. We use the "busy" and "idle" to express the two states. The state of the object can be obtained by the corresponding relationship between objects and the relative positions of two types of objects in the distribution system. For example, in a route, when a vehicle is between a gas station and the last served gas station, the state of the current gas station is "unserved", and the state of the last served gas station is “served”, and the state of the road between the two gas stations is "being traveled". When a vehicle is located in a gas station, the state of the gas station is "being served" and the state of the road between this station and the last served station is "traveled". When a vehicle returns to the distribution center after serving the last station, the state of the vehicle will be from "busy" to "idle". When a vehicle travels from the distribution center to the first station, the state of the vehicle will be from "idle" to "busy". Therefore, in the ontology, in order to capture the state of the object, we need to represent the corresponding relationship between the three objects: gas stations, roads and vehicle. Besides, we need to represent the location information of each object. The objects-relationship ontology is shown in Fig. 3.
1162
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
Fig. 3 The objects-relationship ontology
Specifically, "serve" represents that the relationship between the object "vehicle" and the object "GasStation", which means that the gas station will be served by the vehicle and the vehicle serves all the stations with the same vehicle ID. "belong" represents the relationship between the object "road" and the object "route" , which means that a road is part of a route and the roads which have the same route ID belong to the same route. "travel" represents the relationship between the object "vehicle" and the object "route" , which is a pair relationship. It means that a vehicle can only travels one specific route. "link" represents the relationship between the object "road" and the object "gas station", which means that a road links two gas stations. The object "road" refers to the shortest path between two stations, and the object "route" is a collection of "road". "GSLocation" "RLocation" and "Vlocation" represent the positions of gas stations, roads and vehicles respectively. 3.3. SWRL Rules The knowledge to judge whether the original distribution plan is disrupted by an unexpected event under the current context comes from the experienced scheduler. Such knowledge can be represented as a series of rules in the form of "IF Condition THEN Action". Thus, we use SWRL rules which combine rules and ontology to represent such knowledge. By SWRL, we can use the vocabularies and the relationships depicted in ontology to describe rules. Such rules are called Disruption Analysis and Judgment Rules. According to the characteristics of the unexpected event we can divide the Disruption Analysis and Judgment Rules into six types. For example, we call the rules which are used to analyze the change of demand as "Demand-Change-Judging Rule", which is a rule set containing six rules shown in Table 1. The other five types of rules are "Time-Window-Change-Judging Rule", "Disabled-Road-Judging Rule", "Road-Time-Delay-Judging Rule", "Disabled-Vehicle-Judging Rule", "ReparableVehicle-Judging Rule".
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
Table 1 "Demand-Change-Judging Rule" Rule
Rule Content
Specific Meaning
Rule1
GasStationState (? x ) ġ hasState (? x, served ) ġ UnexpectedEvent(?y) ġ hasEvent (? y, DemandChange)ĺinfluence (? y, disturbsystem)
If the gas station changes the demand at the time when the gas station’s state is "served", the unexpected event will cause a disruption to the distribution system.
Rule2 GasStationState (? x ) ġ hasState (? x, being-served ) ġ UnexpectedEvent(?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: moreThan (?z, 0) ġ swrlb: moreThan (?z, rl) ĺinfluence (? y, disturbsystem) GasStationState (? x ) ġ hasState (? x, unserved ) ġ UnexpectedEvent(?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: moreThan (?z, 0) ġ swrlb: moreThan (?z, rl) ĺinfluence (? y, disturbsystem)GasStationState (? x ) ġ hasState (? x, unserved ) ġ UnexpectedEvent(?y) ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: moreThan (?z, 0) ġswrlb: moreThan (?z, rl) ĺinfluence (? y, disturbsystem) Rule3 GasStationState (? x ) ġ hasState (? x, being-served) ġ UnexpectedEvent(?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: moreThan (?z, 0) ġ swrlb: lessThan (?z, rl) ĺinfluence (? y, nodisturbsystem) GasStationState (? x ) ġ hasState (? x, unserved) ġ UnexpectedEvent(?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: moreThan (?z, 0) ġ swrlb: lessThan (?z, rl) ĺinfluence (? y, nodisturbsystem)
Rule4 GasStationState (? x ) ġ hasState (? x, being- served ) ġ UnexpectedEvent (?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: lessThan (?z, 0) ġ swrlb: sameAs (?z, d) ėinfluence (? y, disturbsystem) GasStationState (? x ) ġ hasState (? x, unserved ) ġ UnexpectedEvent (?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: lessThan (?z, 0) ġ swrlb: sameAs (?z, d) ėinfluence (? y, disturbsystem) Rule5 GasStationState (? x ) ġ hasState (? x, being- served ) ġ UnexpectedEvent (?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: lessThan (?z, 0) ġ swrlb: lessThan (?z, d) ėinfluence (? y, nodisturbsystem) GasStationState (? x ) ġ hasState (? x, unserved ) ġ UnexpectedEvent (?y)ġ hasEvent (? y, Demand-Change) ġ hasEventnum(?y, ?z) ġ swrlb: lessThan (?z, 0) ġ swrlb: lessThan (?z, d) ėinfluence (? y, nodisturbsystem)
If the gas station increases the demand and the increased demand exceeds the redundant load of the vehicle at the time when the gas station’s state is "unserved" or "being served", the unexpected event will cause a disruption to the distribution system.
If the gas station increases the demand and the increased demand is less than the redundant load of the vehicle at the time when the gas station’s state is "unserved" or "being served", the unexpected event will not cause a disruption to the distribution system. If the gas station reduces the demand and the reduced demand equals to the quantity of the order at the time when the gas station’s state is "unserved" or "being served", the unexpected event will cause a disruption to the distribution system. If the gas station reduces the demand and the reduced demand is less than the quantity of the order at the time when the gas station’s state is "unserved" or "being served", the unexpected event will not cause a disruption to the distribution system.
1163
1164
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
4. Conclusion During the oil products distribution process, unexpected events are diverse and can occur at different distribution states and the distribution state changes constantly, which make it hard to capture the distribution context in real time. In this paper, an ontology-based model is developed, which build the disruption ontology, the distribution-state ontology and the SWRL rules to capture the typical-context in real time. The ontology-based model can capture the typical-context in real time and judge whether the original distribution plan is disrupted by the unexpected event under the current context, which is in favor of taking measures to respond to the disruption in real time. The model in this paper could just judge whether the original distribution plan is disrupted by an unexpected event under the current context. The policies responding to the disruptions in real time has not been considered, which is important for recovering the distribution. Therefore, further research should focus on the building of the policies. Acknowledgements This work has been partially supported by the grants from Natural Science Foundation of China (Nos. 71201014, 71571027, 71272093, 71271037); The Fundamental Research Funds for the Central Universities (Nos. DUT14QY28ˈDUT14RC(4)04)
References 1. Potvin JY, Xu Y, Benyahia I. Vehicle routing and scheduling with dynamic travel times. Computers & Operations Research 2006;4:11291137. 2. Li JQ, Mirchandani PB, Borenstein D. Real-time vehicle rerouting problems with time windows. European Journal of Operational Research 2009;3:711-727. 3. Li JQ, Mirchandani PB, Borenstein D. A Lagrangian heuristic for the real-time vehicle rescheduling problem. Transportation Research Part E 2009;3:419-433. 4. Li JQ, Borenstein D, Mirchandani PB. A decision support system for the single-depot vehicle rescheduling problem. Computers & Operations Research 2007;4:1008-1032 5. Ichoua S, Gendreau M, Potvin JY. Diversion issues in real-time vehicle dispatching. Transportation Science 2000;4:426-438. 6. Wang MC, Gao CX, Zeng YT. Recovery of the VRPTW disruption and the TABU SEARCH algorithm. Journal of Mathematics 2006;2:231236. (in Chinese) 7. Song JW, Rong G. Study of uncertainty problem in vehicles scheduling. Journal of Zhejiang University (Engineering Science) 2003;2:243248.(in Chinese) 8. Ding QL. A model of disruption management with the changes of time windows-based on the perspective of behavior. Chinese Journal of Management Science 2015;5:89-97. (in Chinese) 9. Hu XP, Yu N, Ding QL. Sequential decision methods for disruption management in distribution. Journal of Industrial Engineering /Engineering Management 2011;2:186-190, 130. (in Chinese) 10. Wang XP, Ruan JH, Zhang K, et al. Study on combinational disruption management for vehicle routing problem with fuzzy time windows. Journal of Management Sciences in China 2011;6:2-15. (in Chinese) 11. Hu XP, Sun LJ, Wang YN. A model for disruption management in urban distribution systems. Journal of Management Sciences in China 2011;1:50-60. (in Chinese) 12. Wang Z, Hu XP, Wang XP. Disruption management model and algorithm for distribution vehicle scheduling problems under accidental travel time delay. Systems Engineering-Theory & Practice 2013;2:378-387. (in Chinese) 13. Hu XP, Sun LJ, Liu LL. A PAM approach to handling disruptions in real-time vehicle routing problems. Decision Support Systems 2013;3:1380-1393. 14. Sun LJ, Hu XP. A Knowledge Representation Method for Algorithms in DSS for Real-time Vehicle Routing in Urban Distribution. International Journal of Innovative Computing, Information and Control 2012;8:5859-5872. 15. Ye J, et al. Semantic web technologies in pervasive computing: A survey and research roadmap. Pervasive and Mobile Computing 2015;23:125. 16. Allen JF. Towards a general theory of action and time. Artificial Intelligence 1984;2:123-154. 17. Hobbs JR, Pan F. An ontology of time for the semantic web. ACM Trans. Asian Lang. Inf. Process 2004;1:66-85. 18. Lenat DB, Guha RV. Cycling paper reviews-response. Artificial Intelligence 1993;1:149-174. 19. Borgo S, Guarino N, Masolo C. A pointless theory of space based on strong connection and congruence. In: Aiello LC, Doyle J, Shapiro S, editors. Proceedings of the fifth international conference. San Mateo: Morgan kaufmann; 1996. p. 220-229.
Li jun Sun et al. / Procedia Computer Science 96 (2016) 1156 – 1165
20. Kagal L, Finin T, Joshi A. A policy based approach to security for the semantic web. In: Fensel D, Sycara K, Mylopoulos J, editors. Proceedings of the second International Semantic Web Conference. Berlin: Springer; 2003. p. 402-418. 21. Strang T, Linnhoff-Popien C, Frank K. Applications of a context ontology language. In: Begusic D, Rozic N, editors. Proceedings of the International Conference on Software. Croatia: Univ. Split; 2003. p. 14-18. 22. Kabir M, Han J, Yu J, Colman A. Scims: a social context information management system for socially-aware applications. In: Ralyte J, Brinkkemper S, Wryczr S, editors. Proceedings of 24th International Conference. Berlin: Springer; 2012. p. 301-317. 23. Hepp M. Goodrelations: An ontology for describing products and services offers on the web. In: Gangemi A, Euzenat J, editors. Proceedings of the 16th International Conference on Knowledge Engineering: Practice and Patterns. Berlin: Springer; 2008. p. 329-346. 24. Yan Z, Chakraborty D, Parent C, Spaccapietra S, Aberer K. Semantic trajectories: mobility data computation and annotation. ACM Trans. Intell. Syst. Technol 2013;3:1-38. 25. Spaccapietra S, Parent C, Damiani ML, DeMacedo JA, Porto F, Vangenot C. A conceptual view on trajectories. Data Knowl. Eng. 2008;1:126-146. 26. Chen H, Perich F, Finin T, Joshi A. SOUPA: standard ontology for ubiquitous and pervasive applications. In: ieee computer society, editors. Proceedings of Mobiquitous 2004. Los Alamitos: IEEE Computer; 2004. p. 258-267. 27. McDermott D. A temporal logic for reasoning about process and plans. Cognitive Science 1982;2:101-155. 28. Allen JF. Towards a general theory of action and time. Artificial Intelligence 1984;2:123-154. 29. Qu YZ. A predicate-ordered logic for knowledge representation on the web. Future Generation Computer Systems 2004;1:19-26. 30. Riccardo R. On the decidability and complexity of integrating ontologies and rules. Web Semantics: Science, Services and Agents on the World Wide Web 2005;1:61-73. 31. Lin X, Liu WY. Context based knowledge representation and reasoning-from the perspective of artificial intelligence. Computer Science 2005;1:142-146. (in Chinese) 32. Li HY, Wang YB, Qu YZ, Pan JZ. A Reasoning Algorithm for pD*. In: Riichiro M, Zhong ZS, Fausto G, editors. Proceedings of First Asian Semantic Web Conference. Berlin: Springer; 2006. p. 293-299. 33. Wen KM. Research on Semantic Search Based on Ontology Repository Reasoning. Wuhan: Huazhong University of Science and Technology, 2007. (in Chinese) 34. Badder F, Sattler U. Tableau Algorithms for Description Logics. In: Dyckhoff R, editors. Lecture Notes in Artificial Intelligence. Berlin: Springer; 2000. p. 1-18. 35. Horrocks I, Sattler U. A Tableaux decision procedure for SHOIQ. Journal of automated reasoning 2007;3:249-276. 36. Peng S. Research on ontology reasoning based on description logic in semantic web. JiLin: Jilin University, 2009. (in Chinese) 37. Badder F, Sattler U. An overview of tableau algorithms for description logics. Studia Logica 2001;1:5-40. 38. Horrocks I, Patel-Schneider PF. Evaluating optimized decision procedures for propositional modal Kappa((m)) satisfiability. Journal of automated reasoning 2002;2:173-204. 39. ter Horst HJ. Extending the RDFS Entailment Lemma. In: McIlraith SA, Plexousakis D, VanHarmelen F, editors. Proceedings of 3rd Int. Semantic Web Conference. Berlin: Springer; 2004. p. 77-91. 40. Motik B, Sattler U, Studer R. Query answering for OWL-DL with rules. J. Web Semantic 2005;1:41-60.
1165