Recommendation reason generation method and device, electronic equipment and readable storage medium

Abstract

The invention discloses a recommendation reason generation method, which comprises the steps of obtaining at least one recall result for search data according to the search data of a target user; according to the search data of the target user, the recall results and the portrait of the target user, obtaining recommendation reasons of the recall results through a preset intelligent question and answer model; wherein the intelligent question and answer model is a first machine learning model obtained by training at least one sample question and answer data combination; wherein the sample question and answer data combination comprises a sample user portrait and historical evaluation data of at least one sample user, a recommendation object corresponding to the historical evaluation data, andhistorical search data of the sample user for the recommendation object. The technical problems of high cost and insufficient individuation are solved. The method and the device have the beneficial effects of reducing the recommendation reason generation cost and improving the individuation degree of the recommendation reason.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a method, device, electronic equipment and readable storage medium for generating recommendation reasons. Background technique [0002] With the development of AI (Artificial Intelligence, artificial intelligence) technology, related applications are generating more and more important commercial value and social influence, how to solve the trust mechanism in its decision-making process is the key to promoting the further development of artificial intelligence factor. For example, for a search system, the subject of evaluation is the user it faces, and the user is highly subjective, so the interpretability of the results will not only directly affect the effect of the search system, but also affect the user's trust and confidence in the system. Acceptance. In recent years, the explainable search system has attracted more and more attention. It can not only improve the ...

AI Technical Summary

Problems solved by technology

[0004] However, the above-mentioned four mainstream methods for generating recommendation reasons have certain defects and limitations.

Among them, for the manual operation method, although the manually written sentences are of high quality and rich in content, the cost is high, the number is limited and the update is slow, and it cannot meet the personalized generated content; for the rule template method, compared with manual operation, it can reduce certain costs , but the content is monotonous, the coverage of the dimensions that can be introduced is low, and it is not universal, and it cannot meet the needs of individualization; the method of extracting comment data is strictly dependent on the supply of comment data, which has certain limitations; for the content generation method, First of all, in the scenario of recommendation reasons, there is a lack of good samples, and secondly, for a single merchant, it cannot satisfy personalized generated content

It can be seen that the existing recommendation reason generation schemes have technical problems such as high cost and insufficient personalization.

Images