Design And Implementation Of Report System For Semantic Query

In the era of Internet,the most obvious feature is the increase of information,which lead to the process of access to information more difficult.Traditional information retrieval techniques are based on keyword matching,which is a kind of technique of partial substitution,and cannot understand user's true intend,which results in the low precision ratio and low recall ration of retrieval information.In the time,the semantic retrieval technique based on semantic matching is born.At the request of an electric company,this paper designs and implements a reporting system that supports semantic query.The report system has the functions of report design and production,personalized recommendation,authority management,and semantic retrieval.Which focuses on the design and implementation of the semantic retrieval module,for semantic retrieval with the report existing in the system,allows users to quickly find what they want.In order to realize the semantic matching,firstly we get terminology and relationship from Birt that is used for building report,then use protege to create ontology,and use the knowledge network to create a semantic knowledge base in the field of report.For report file,based on the report ontology,using Jena to extract parameter and semantic extension,then using Lucene to establish index structure including parameter field and semantic information field.The semantic extension of report information model is the semantic triples.The query that user enters is processed into the same structure as the index structure.At first,we using LTP to segment and part of speech tagging the words of user entering,in the aftermath of the word segmentation and part of speech tagging to extract the user's retrieval condition of statements,then for user's retrieval conditions and user's input words according to the report ontology using Jena to synonym expensive and semantic expensive.The retrieval conditions for the extraction,we use Lucene query function to precision matching of the parameter fields established in the index structure.As the extension result,we use Lucene query function to fuzzy matching,resulting in the semantic retrieval.During the implementation phase,this paper shows the semantic knowledge base of the report domain,the report ontology,and the retrieval system of semantic matching.Contrast of the query results from query system based on keyword matching and the results from semantic query system,statistical precision rate and recall rate,and make a scatter diagram,show that the semantic matching based on ontology retrieval system is more suitable for the requirements of an electric company than keyword matching retrieval system.