Context-aware Information Retrieval In Internet Of Things

With the rapid development of technologies such as the sensor, communication, computer, network, and so on, traditional internet is developing towards the more extensive Internet of Things (IOT). There will be billions of objects connected to IOT, and they require communication and information exchange with each other instantly and frequently. However, the data in IOT is huge, various, dynamic, real-time, semantic rich and heterogeneous, traditional technology of data management and information retrieval has failed to meet the search needs in IOT, as one of the major bottlenecks to restrain the current development of IOT.The future search in IOT will not only supply users a series of document links, but also provide the answers of users questions according to objects and users current context information. Context-aware information retrieval is intelligent information retrieval technology emerged in recent years, which submit the best appropriate results to users enable the user to get the experience of "what you get is what you want" according to the state of users and entities in the environment (including the user's own information and information about the surrounding environment). Meanwhile, service-oriented architecture has been shown especially to be suitable for the management of information in IOT. There will be more and more services to be published to the IOT, and the user can get various information through the services. In this paper the basic characteristics of the data in IOT are analyzed, then a context-aware and service-oriented information retrieval architecture for IOT was designed. Moreover, there are three kinds of technologies are addressedd in this paper. They are context information management, context-aware semantic search and search results ranking. In this architecture, context information includes static and dynamic information mainly acquired from web page, database, UDDI and so on. Context ontology is used to model context information; a number of SWRL inference rules and SPARQL queries are defined to achieve semantic retrieval results for user queries. In the end, a context-aware ranking algorithm is presented to rank the search results.Based on the above architecture, a recommendation system for student dining is designed and implemented. The system can recommend the reasonable diet and appropriate diningroom to the student according to the context information of student (including the user's preferences, current location, the recent dining records, medical records, etc.) and diningroom (including the current menu, the remaining reception capacity, congestion factor and so on). The system can answer the students "what to eat today", "where to eat" and other questions which traditional information retrieval system can not answer. Experiments have shown that the constructed context-aware model of information retrieval in IOT is reasonable and effective.