Research On Queries' Time Attribute And Its Application On Ranking Of Retrieval Results

With the rapid expansion of information, information retrieval system has become the main tool of information searching. How to make information retrieval system understood user's search intention, found the information that satisfy user's requirement, and presented good sort results to user have become hot research topics. Queries'attribute analysis is an important resource of getting user's search intention. In this paper, queries'time attribute is studied and applied on ranking of retrieval results.In this paper, an information retrieval system based on queries'time attribute is built which is constructed by three functional modules including query classifier, time attribute recommend system and rank module. First, SVM technology is used to built the classifier to automatic identify a class of queries that have implicit time attribute, then analyzing distributions of length and category of those queries which have implicit time attribute. Furthermore, a time attribute recommend system is constructed by building different time attribute query set's language model, and is used to recommend time attribute to those queries that are identified have implicit time attribute by comparing every language model's probability. Finally, a new rank module which introduce queries'time attribute into search results sorting is built, based on the theory that word does not distribute uniformly in different time attribute query set, so the word's importance should vary with time attribute, in different time attribute, word is given different weight.Three experiments are built to test classifier, recommend system and the new rank module's performance. And the results of these experiments show that the information retrieval system's search results can be better to satisfy user's search demand by taking queries'time attribute into account.