Model acquisition method, keyword generation method and device, medium and computing equipment

Abstract

The embodiment of the invention provides a model acquisition method, a keyword generation method and device, a medium and computing equipment. The model acquisition method comprises the following steps: a keyword generation model is constructed, wherein the model comprises a main module based on a coder-attention mechanism-decoder framework; in the supervised learning stage, for each training datapair, the main module constructs probability distribution based on semantics and domain information of a source keyword and a reference target keyword in the training data pair, samples a first hidden variable from the probability distribution, and generates a first target keyword; parameters of the main module are updated in a gradient back propagation mode by minimizing a loss function of the main module. According to the technology, low-frequency related keywords except hot high-frequency keywords can be generated, and the generated target keywords are more related to the field of the source keywords by introducing field constraints; in addition, the correlation and domain consistency of the keywords can be further improved through reinforcement learning.

Description

technical field [0001] Embodiments of the present invention relate to the field of electronic information, and more specifically, embodiments of the present invention relate to a method for acquiring a model, a method for generating keywords, a device, a medium, and a computing device. Background technique [0002] With the development of science and technology, electronic digital devices usually have many functions such as website browsing and running mobile client applications. In addition, there are terminal devices specially used for e-book reading, such as electronic paper books. Among the functions mentioned above, some are equipped with search engines, and users can input query words in the search box, and then search for relevant content. [0003] Taking commercial search as an example, there are usually two issues to consider: first, the keyword bidding problem, popular advertising keywords are very expensive, this is because more advertisers will search on keywords...

AI Technical Summary

Problems solved by technology

[0003] Taking commercial search as an example, there are usually two issues to consider: first, the keyword bidding problem, popular advertising keywords are very expensive, this is because more advertisers will search on keywords with more popular bids, and Unpopular keywords are difficult to find, so most ads have little chance to be shown to users; secondly, the problem of inefficiency, in a large percentage of search queries, keywords that are relevant to many ads but are not popular appear in search No ad pages were shown in the results, and existing retrieval- or match-based approaches either worsened bidding or failed to come up with solutions that would cover more queries, resulting in inefficiencies and fewer ad impressions

Images