Research On Retrieval Technology Based On User Information In Internet Of Things

With the rapid development of Internet of Things, all kinds of sensor deviceswill be embedded in Internet of Things to perceive all objects and infrastructure in thenetwork. However, the differences between different categories of sensor often leadto large differences in data forms and details, and this may make information morecomplex in the Internet of Things. In the face of growing heterogeneous data ofInternet of Things, the main problems need to be solved for Internet of Thingsretrieval technology is how to use effectively and acquire the useful information forusers themselves.The traditional retrieval systems because of single retrieval form, could notsatisfy heterogeneous data and real-time characteristics of the complex applicationenvironment in Internet of Things, could not satisfy the needs of user for data typesand retrieval accuracy,thus it will bring poor extensibility, single feedback form, highenergy consumption and load unbalancing. According to the huge amounts ofheterogeneous data of Internet of Things and the needs of user, combining with thearchitecture of Internet of Things, this thesis presents a retrieval algorithm based onusers information in Internet of Things which make the retrieval system think interms of the user`s needs and interests, enables get effective retrieval and unifiedmanagement for the heterogeneous data in Internet of Things, provides retrievalservices for users. The mainly research tasks of thesis are as follows:Firstly, the thesis presents a heterogeneous data preprocessing algorithm, andrealizes the unified management for Internet of Things by using three levels ofpretreatment model. Uses hash function for heterogeneous data attribute featureextraction, and maps to the information service node, builds the mapping informationchart and routing turn table, uses the Chord algorithm to provide retrieval service forthe users.Secondly, in order to achieve the useful resources for users in huge amounts ofinformation, the thesis introduces the idea of collaborative filtering recommendation algorithm, the users could grade the retrieval resource; the system analyzes thepreference information provided by user registering information and retrievalbehavior, this thesis gives full consideration to the similarity between users andretrieval projects, and regards the users and retrieval resources as a cooperativeretrieval network. According to the neighbor of the user, the retrieval system couldfeed back high similar recommendation results.Finally, in side of network delay, single point failure retrieval efficiency, andrecall, using the simulation to analyze the algorithm proposed by thesis, making fulluse of the huge amounts of data, in the same time of ensuring the independence ofheterogeneous data, realizing the retrieval service based on users information.