Method and device for automatically verifying commodity categories

Abstract

The invention discloses a method and device for automatically verifying commodity categories, and relates to the technical field of computers. A specific embodiment of the method comprises the steps of obtaining titles and actual categories of commodities; calculating titles of the commodities by utilizing an ith-level classification model, and determining the probability that the commodities belong to each ith-level category in an ith-level category set, wherein i is an integer greater than or equal to 1 and less than or equal to N, and N is an integer greater than or equal to 1; from the first-level category set, selecting the first-level category with the maximum probability as a first-level prediction category; selecting a kth-level category containing (k-1)-level prediction categoriesas a kth-level prediction category from a kth-level category set, and taking the kth-level prediction category as a prediction category of the commodity, wherein the k is an integer and k is greaterthan 1 and less than or equal to i; and verifying the actual category according to the prediction category. According to the embodiment, text semantics and upper-layer categories can be comprehensively considered, the commodity categories are automatically verified, the labor workload is reduced, the accuracy and efficiency are improved, and the expandability is high.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a method and device for automatically checking commodity categories. Background technique [0002] The commodity category system is one of the most important infrastructures in the field of e-commerce, and its quality will directly affect the application of subsequent search recommendations, intent recognition, and related personalized services. However, during the operation of the actual e-commerce website, the phenomenon of wrong listing of categories often occurs, which leads to products being mistakenly classified under other irrelevant categories, which reduces user experience and product conversion rate. At present, there are two main ways to detect whether the category is not linked. One is to rely on the operator to determine whether the product title and category of the merchant are correct, and the other is to filter out the products whose title does not match the con...

AI Technical Summary

Problems solved by technology

[0004] However, for Internet e-commerce companies, hundreds of thousands of new products are added every day. It takes a lot of manpower and time to check whether the products match the category manually, and the cost is high; the biggest disadvantage of the second method is the limitation of the rules And non-extensible, the core of the rules is to judge by keywords, the fewer the rules, the phenomenon of many wrong listings is not easy to be detected, the fewer the rules, the greater the impact on the performance of the algorithm, and the probability of conflicting rules Larger, less accurate

To sum up, if only relying on human work or existing rule review, it is far from meeting the need to review category errors from hundreds of millions of pieces of data

Images