Method and system for identifying core product words

Abstract

The invention discloses a method and system for identifying core product words, and relates to the technical field of computers. The method includes the following specific steps of obtaining a displayimage of a product, determining a plurality of candidate product words contained in a title of the product, and determining a plurality of product image sets, wherein each product image set in the plurality of product image sets and each candidate product word in the plurality of candidate product words corresponds to each one; for each product image set in the plurality of product image sets, determining the similarity of the product in the display image and.the candidate product words corresponding to the product image sets according to the display image and each image in the product imagesets to obtain multiple similarities; and determining the candidate product words corresponding to the similarity greater than a preset threshold in the multiple similarities as the core product words. This method adds image information to the recognition of the core product words, so that the core product words of the product title is determined more intuitively and accurately.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a method and system for identifying core product words. Background technique [0002] In the field of e-commerce, accurate analysis of the composition of product titles is the basis for user intent recognition, product recall, and personalized recommendations. Different from general natural language analysis, in e-commerce scenarios, it is necessary to distinguish brand words, modifier words, product words and other components in product titles. Moreover, in the field of e-commerce, many product sellers will pile up and list product words in the title of the product in order to increase the click volume of the product, and even many of the product words do not correspond to the product. After the seller puts the product on the shelf, in order to accurately provide the product to the buyer, it is necessary to identify the core product words in the product title. Among them, th...

AI Technical Summary

Problems solved by technology

[0004] However, the methods for identifying core product words in commodity titles in the prior art only utilize text information

Specifically, the vocabulary method does not have the ability to generalize, that is, it cannot handle new samples that are not included in the vocabulary, and there are many types of commodity titles, it is impossible to include all cases into the vocabulary

In the e-commerce scenario, the phenomenon of product words in the title is serious and it is difficult to use syntactic information to analyze, so the method of machine learning cannot be very effective in identifying the core product words in the product title.

Moreover, the various and irregular title writing forms of merchants make it difficult to judge the core product words in the title simply by using text information.

Images