Font Size: a A A

Research And Implementation Techniques Of Information Retrieval Based On User Query Intention

Posted on:2014-08-04Degree:MasterType:Thesis
Country:ChinaCandidate:M ZhangFull Text:PDF
GTID:2268330425486761Subject:Software engineering
Abstract/Summary:PDF Full Text Request
With the rapid development of Internet, search engine plays a more and more important role in the application of the Internet, how to search engine to obtain the information they need has become the face of every Internet user moment problem using. The book as the main approach for people to obtain knowledge and information that have been gradually replaced by the network, in this case, the digital library has become the countries to actively promote the project.As the core force of Chinese Digital Library of CADAL has tens of thousands of copies of the digital resources, in order to facilitate users to get the books they want, CADAL provides the traditional retrieval system responsive to user. But in the practical application, users often with some uncertainty, in this case, the traditional retrieval system and user retrieval intention returned inconsistent results, so it is necessary to improve the retrieval system, to improve the retrieval efficiency of users, so that digital resources can be used to the greatest extent.This paper is for the current lack of search engine, the user query information retrieval technology based on intention, combined with the characteristics of CADAL library resources at present, mining system log CADAL platform produced by the analysis of potential users, intention, so as to obtain the query-potentially graph mapping and potential intentions-Book Mapping Using Lucene, a full-text retrieval system, provides a more efficient and accurate search experience for the user.This paper first discusses some key technique in information retrieval, describes the related theory of information retrieval, introduced emphatically the Lucene open-source full-text search related knowledge framework and personalized recommendation, and in accordance with the recommendation of personalized recommendation to do the classification, the method of matrix decomposition are analyzed and explained in detail.The paper then discusses the basic concepts of data mining, the main data mining algorithms, discusses in detail the design thought of potential users of SVD model mining algorithm based on intention and process. It introduces the whole architecture of CADAL platform, analysis of the nginx log format, for each field log illustrates the meaning. The access to the mapping relationship between query words-books from the log file, by using the singular value decomposition of SVD the query-potential intentions mapping relationship with potential intentions-Book mapping.Then the paper describes the design object retrieval system based on click log analysis, give the architecture of the system design, design the overall function structure diagram of the system, the log processing, intention, function mining indexing and user interface, the key of the detailed design, design the conceptual model of database, describe method the establishment of a data set table.Finally, from the perspective of concrete realization describes the overall system architecture, this paper discussed the basic development process of Lucene, the results retrieved from the launch of potential query the intentions of the user process was extremely detailed analysis, through the analysis of user behavior log information, and decomposition technology potential user query intention from the system log mining history with SVD matrix, and the use of retrieval technology, personalized recommendation technology, combined with the Lucene open source framework, design and implementation of user query intention retrieval system based on user, through the acquisition to the actual query intent of these log analysis, the analysis of query intent can greatly improve the retrieval efficiency of the user in the retrieval process.
Keywords/Search Tags:Information Retrieval, Digital Library, Personalized Recommendation
PDF Full Text Request
Related items