Statement recommendation method and device based on associated interest points

Abstract

The invention provides a statement recommendation method and device based on associated interest points, wherein the method comprises the steps of obtaining a statement inputted by a user; extractinga keyword in the inputted statement, and searching a current interest point matched with the keyword in a preset interest point association network; determining a plurality of related interest pointsmatched with the current interest point according to a preset interest point association network, and screening the plurality of related interest points according to a preset screening strategy to obtain the extended interest points; and retrieving the first reply statement and the second reply statement in a preset corpus according to the current interest point and the extended interest point, and providing the first reply statement and the second reply statement for the user. Therefore, the expanded interest points are guided through the association between the different interest points, then the reply sentences containing the expanded interest points are provided for the user, and the user interest point set is expanded based on the user feedback, the user interest portrait can be expanded more efficiently, the topic diversity in the man-machine conversation is improved, and the user experience is improved.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a sentence recommendation method and device based on associated interest points. Background technique [0002] At present, in the human-machine dialogue scenario, chatbots usually use the user's voice information as input, and automatically decide the semantic content of the reply by analyzing the voice information, combining the previous dialogue history of the two parties, user interest information, etc. [0003] However, the initial user interest information is usually specified by the user, or comes from a small number of brief interactions with the user, usually the granularity of the information is not fine enough (for example, only at the level of specific fields: entertainment and sports), and the coverage is not enough, and it usually relies on the active guidance of the user Find a new topic, and then extract user interest information based on th...

AI Technical Summary

Problems solved by technology

[0003] However, the initial user interest information is usually specified by the user, or comes from a small number of brief interactions with the user, usually the granularity of the information is not fine enough (for example, only at the level of specific fields: entertainment and sports), and the coverage is not enough, and it usually relies on the active guidance of the user Find a new topic, and then extract user interest information based on the content of the user's conversation. Because it is completely dependent on the user to start the topic, it is easy to fall into a relatively closed topic collection, and the accumulation process is relatively inefficient, resulting in a relatively low coverage of user interest.

Images