Question answering system-based information matching method and system

Abstract

The invention discloses a question answering system-based information matching method and system. The method comprises the following steps of: configuring user interactive behavior representation parameters for question answering pairs in a question answering system knowledge base; dynamically updating the user interactive behavior representation parameters according to user interactive behavior feedback information, wherein the user interactive behavior feedback information is feedback which responds to question information of the question answering system; acquiring a question candidate set in the question answering system knowledge base according to question information input by a user; sorting all pieces of question information in the question candidate set by taking the user interactive behavior representation parameters of all pieces of question information in the question candidate set as an index; and returning a predetermined number of pieces of question information which are sorted on the front side to the user. According to the method and the system, information matching speed and accuracy of the question answering system can be improved.

Description

technical field [0001] The invention relates to search engine technology, in particular to an information matching method and system based on a question answering system. Background technique [0002] Studies have shown that in the face of massive user information on the Internet, most users use search engines as the starting point for accessing the Internet to obtain information. Traditional search engines are mainly based on webpage search, and most of them use keyword matching to return a list of webpages containing the searched keywords, which are viewed by users after clicking on the webpage. With the continuous development of science and technology, handheld devices such as mobile phones and PDAs (Personal Data Assistants) are gradually popularized, and more and more users access the Internet through handheld devices, and the disadvantages of traditional search engines are also beginning to appear. The screen size and communication efficiency of handheld devices limit...

AI Technical Summary

Problems solved by technology

Since the questions of the question answering system are often relatively short, the word frequency information is relatively concentrated and less in the question sentences, mostly 1-2 times, so it is difficult to reflect the meaning contained in the word frequency information of TF-IDF

[0006] Sorting methods using semantic matching (including vector space methods using semantic information to calculate distances) generally need to traverse the candidate set, which has a problem of high time complexity

Moreover, the quality of semantic distance information is not high so far, and the reliability and coverage limit the scope of application.

[0007] The sorting method based on the matching degree of grammatical information is often a supplement to the first two methods due to the flexibility of Chinese and the looseness of grammar. It extracts 1-2 obvious grammatical features for computing sorting, and cannot be used for sorting alone method, and there are also problems of complex calculation and low precision, which will directly affect the sorting results

[0008] To sum up, in the prior art, the information matching in the question answering system, specifically, the matching method of question sentence information still has many defects, and the matching accuracy and matching efficiency need to be improved

Images