Optimize And Improve Semantic-based Full-text Search

With the rapid development of internet technology，information explosion brings peoplehuge information resources. The amount of information increases sharply. informationacquisition becomes more and more difficult. It has become an important reason thatrestricts the development of the information. The existing full-text retrieval technology alsohad different problems in the face of retrieval requirements of different areas. People aboutRequirements of information acquisition is getting higher and higher, today People to obtaininformation is not only to find it. The experience of user is continuous improving.Traditional retrieval technology always care for this and lose that when in the pursuit ofrecall and precision. Semantic technology developed rapidly in recent years, has beenapplied to many fields. This paper is to study the semantic based full-text retrievaltechnology. In the process of system design, the system did not abandon traditional full-textretrieval technology, but development of the combination with Semantic Technology basedon the current retrieval technology.The use of semantic technology in this article is mainlyembodied in the application of ontology. It is used in the pretreatment on query andindexing of documents etc.This paper studies and analyzes the overall framework of the full-text retrieval, on thebasis of the former excellent experiences, targeted improvement and optimization isproposed for the full-text retrieval technology.Firstly this system based on semantics, in thepreprocessing stage of query using ontology technology, by using the Domain Ontologyknowledge base at the semantic level query words are expanded. Through calculatingsimilarity retrieval feature words one by one and the corresponding ontology tree node toget the extended set. Taking advantage of the expanded query retrieval, the results giveconsideration to recall ratio and precision ratio. In the retrieval phase, It studies the queryresult set output ranking algorithm。The article designs the feature vector model and thecandidate document scoring model to compensate for defects which from single mode. Inanother hot topic retrieval technology, application of query recommendation is alsoappeared in the study of this system in the form of the relevant search which is the mostcommon application of query recommendation in retrieval technology. This paper designsthe related search model based on user log. The query relation which is spontaneous built byuser has the relativity associated most with practical application. He related search extended model based on the user query search logs than any quantitative method to establish modelhas more practical significance. This paper use session markers to analyze user log and putforward user log fragmentation process, in order to provide users with the best relevantsearch recommendation. Semantic application in the paper is based on Domain Ontology.Establishment of full-text retrieval is almost pointed to specific areas of application. Somany of the technical points are targeted. Experiments show that, through the improvementand optimization of the full-text retrieval technology, improves the retrieval effect.