Ranking Algorithm Based On The Semantic Retrieval Of Lexical Semantic Tree

Based on current IR situation that retrieval performances of most IR systems aren't high and the research for semantic retrieval system is conducted enough, the paper proposes a new semantic retrieval model, its core is ranking algorithm for semantic relevance, in which, the set of keywords of text represents text context, general converged extent or aggregate degree of a word and all keywords of text represent semantic relevance of it to text context. Because meaning of text is represented by set of keywords, the integrated result of this semantic relevance of all keywords among sets acquires semantic relevance between texts further. A new mathematic model for semantic computing is designed by using semantic tree (i.e. vocabulary semantic structure) to achieve the computing for these semantic relevant weights. By virtue of query information and document content all being text form, semantic relevant weight of document to query information is acquired by using the mathematic model for semantic computing finally, semantic relevant documents are ranked by their weights, the result of which is user's query result.Grounded on the ranking algorithm model for semantic computing, a retrieval experiment prototype system for English information is built. In implementing it, semantic tree is constructed by using WordNet system that exists as a semantic thesaurus at present, moreover, general framework, programming structures and data structures for the system are designed specially to make actual running efficiency of final implemented system optimal, in this way user can acquire good query experience in querying information.For the implemented system, a lot of deep and detailed tests are done, it is validated by the test results and corresponding analysis that the ranking algorithm for semantic computing is reasonable and correct, simultaneously its good actual retrieval effect makes it have certain actual application value and broad actual application prospect.Some shortages of the system are found by analysis, aiming at the shortages, some future works are put forward, it is hoped that the future works can modify and improve the system to make it more perfect.