Research And Realization Of Intelligent Retrieval Based On Recognition Of Natural Language

The information is blasted in the modern times. More and more, information retrieval is used widely. In fact, information retrieval is that the connotation and extension of querying carry through a match with the document concourse then return the result of querying basing on matching status. But there is much shortage in the traditional information retrieval system. The leak and error in search often happen, so the Recall and Precision is no good. And it is complex to realize information retrieval system, consumers complain about discommodious using.With the research of many retrieval systems, the paper brings forward a new intelligent retrieval model basing on Recognition of Chinese Natural Language for enhancing Recall and Precision. At the same time, realizing the model is brief and good using for the retrieval consumer .The research object is Chinese text. The query based on the full text retrieval of keywords matching has three levels and counts the matching degree corresponding to the level. With the seemly word of emending, the meaning of search is emendated for avoiding error of search. And the word meaning is extended for reducing the leak of search. Then the ratio of the matching degree and word count is regarded as the criterion estimating whether the keywords and the document are similar or not. At last the system exports the result of search with a filtering rule and an order. The intelligent retrieval model can improve Recall and Precision. Then a full retrieval system is realized with the model. Experiments prove the system shows good average Recall and Precision, good using. As follow there is much contribution in paper.1. The paper brings forward a new intelligent retrieval model basing on Recognition of Natural Language. It is easy to realize the query model. And the model gets over much flaw of the traditional information retrieval, obtains good query Recall and Precision.2. The synonymous word-stock system is designed and realized. With the Chinese rule of phoneticizing and multilevel index, a good synonymy system is realized. The systems provide good sustaining for improving query quality.3. A full retrieval system is realized with the model. With laying particular stress on Recall (or Precision) ,the user can set two parameters that is the match rate and the filter amplificatory multiple, and emendate query. It is easy to use the system. The user can directly search all kind of information in document...