Keyword recommending method and device

Abstract

The invention discloses a keyword recommending method and device. The method and device can improve correlation of keywords and published information. The method includes the steps that input information is received; the input information is segmented into a plurality of lexical items; an inverted index structure which is built in advance is inquired by means of each lexical item, and the keywords obtained by inquiring the inverted index structure through all the lexical items form a candidate recommendation word set; the correlation score of each keyword in the candidate recommendation work set and the input information is calculated, and one or more keywords are selected according to the correlation scores to serve as recommendation word to be output. The keyword recommending device comprises a receiving module, a pre-processing module, a recall module and a keyword evaluation module. By means of correlation scoring, one or more keywords which are most correlated are selected to serve as recommendation words according to the correlation scores, and the problem that in the prior art, the keywords are not sufficient in correlation is solved.

Description

technical field [0001] The present invention relates to the field of computers, in particular to a keyword recommendation method and device. Background technique [0002] Online bidding advertisement is a widely existing Internet information recommendation scheme, including information recommendation based on keywords. The keyword-based information recommendation method is to determine the information to be pushed to the user according to the keywords entered by the user in the search engine. When users place information inside and outside the website, they must choose to put in keywords. How to obtain suitable keywords becomes one of the key operations. After users who place information select keywords, they bid on keywords, and the website or information delivery platform formulates delivery and deduction mechanisms. Common deduction mechanisms include CPC (Cost Per Click, deduction based on clicks, the more information is clicked, the greater the deduction), CPM (Cost P...

AI Technical Summary

Problems solved by technology

[0008] Based on the combination of offer terms, the title and attributes of the user's offer are limited by the length, the information contained is generally less, and the quality is uneven. It is difficult to recommend a sufficient number of keywords for the user to choose, and the recall of keywords The rate is obviously insufficient

The IR-based offer keyword recommendation can partially improve the recall rate of keywords, but it cannot recall synonyms and near synonyms, as well as keywords with different literal meanings but related potential meanings.

[0009] 2. Insufficient relevance of keywords

Therefore, the IR-based keyword recommendation method still has the problem of low recommendation accuracy

Images