Method and system for identifying multiple query intents

Abstract

The invention provides a method and a system for identifying multiple query intents. The method includes the steps: calculating probability feature vectors of the query intents according to a G-PLSI model; and calculating similarity between the probability feature vectors of the query intents of different queries, and clustering the queries according to the similarity. The G-PLSI model is used for simulating the generating process of a summary text, searching link clicking behaviors of the different queries with the same query intent and searching behaviors of the different queries with the same query intent in the same session. The probability feature vectors of the query intents reflect summary text information, clicking probability of the different queries on the same link and co-occurrence probability of the different queries in the same session. By the aid of the probability feature vectors of the query intents, search intents of a user can be more accurately reflected by comprehensively using query contents and clicking behaviors of the user.

Description

technical field [0001] The invention relates to the field of information retrieval, in particular to a query multi-intent recognition method and system. Background technique [0002] In the modern age where the amount of information continues to grow rapidly, search engines have become one of the main ways for people to obtain knowledge and useful information. According to the query log information statistics of search engines, the average query length is 2.21 words, of which about 62% are 1 or 2 words in length, and less than 4% are longer than 6 words in query length. Due to the short length of most queries, the search intent expressed by the user in the query often has ambiguity or multiple needs. For example, when the user searches for the word "apple", it may refer to the fruit or the apple. Company, and possibly products of Apple Inc. In addition, because users lack professional knowledge in some fields, it is difficult to use search words to express their meaning cl...

AI Technical Summary

Problems solved by technology

Clustering that comprehensively considers the similarity of content, clicked URL links, and session information includes query clustering proposed by Wen et al., which comprehensively considers the similarity of query content, clicked URL links, and session information similarity. This clustering method only uses a simple weighted form to comprehensively calculate the similarity of querying different information, and cannot identify the user's search intention well.

[0004] In query multi-intent recognition, due to the lack of features of the query text, most of the current related research focuses on clustering based on content similarity or click or session information similarity. These methods lack the consideration of user search intent and cannot be accurate. Distinguishing between the various intents of a multi-intent query

Images