Chinese name self-expanding recognition method based on search logs

Abstract

The invention belongs to the field of natural language processing of computational linguistics and discloses a Chinese name self-extension recognition method based on search logs. In the method, the search logs are utilized to query characteristics of first word family names of strings and excavate seed names by virtue of a 'family name drive' name recognition idea; the seed names are utilized to generate a candidate name template set in the search logs; name templates are screened according to the frequency change tendency of candidate name templates in corresponding query strings and the whole search logs; candidate names are generated according to the name templates; the candidate names are defined and screened by utilizing forward and reverse key words to obtain a name set, previous n templates with highest distinction degree in the name templates serve as seed templates for next iteration, accordingly the names in the search logs are excavated, the seed names are established and the name templates are generated by utilizing characteristics of the search logs, the name templates are filtered according to the change tendency of name contexts in the corresponding search logs and the whole search log query strings, noise information during name recognition is decreased, and the name recognition rate in the search logs is improved.

Description

technical field [0001] The invention belongs to the field of natural language processing in computational linguistics, and in particular relates to a self-expanding recognition method for Chinese personal names based on a search log. Background technique [0002] With the rapid growth of network information, search engines have increasingly demonstrated their epoch-making significance. Since the development of Chinese search engines, they already have a huge number of users, process hundreds of millions of requests every day, and accumulate large-scale query logs. Named entities make up a large proportion of search logs. According to the research statistics of relevant scholars: in the web search queries updated every day, 2~4% of the queries are composed of individual names; about 30% of the queries contain names; the researchers marked 76,717 query strings and found that the names appeared 961 queries were received with a frequency of 6245, accounting for 8.14% of the to...

AI Technical Summary

Problems solved by technology

[0005] In order to solve problems such as low efficiency of name recognition in current search logs, the present invention provides a self-expanding recognition method for Chinese names based on search logs. The method includes the following steps:

Images