User intention identification method and apparatus, electronic device and storage medium

Abstract

Embodiments of the invention provide a user intention identification method and apparatus, an electronic device and a storage medium. The apparatus comprises a determination module used for determining multiple user intentions and multiple feature items, an extraction module used for extracting feature values, corresponding to the user intentions, of the feature items, a first obtaining module used for obtaining user historical behavior data, a training module used for performing model training according to the feature values and the user historical behavior data to obtain weight values of thefeature items, a second obtaining module used for obtaining the current feature items with the real-time feature values, and an intention probability distribution calculation module used for performing weighted summation on the real-time feature values according to the weight values to obtain current probability distribution of the user intentions. The problem of incapability of reflecting an instant tendency of a user under the condition of no query in the prior art is solved, so that the user intentions have comparability.

Description

technical field [0001] The present invention relates to the technical field of the Internet, in particular to a user intention identification method, a user intention identification device, an electronic device and a storage medium. Background technique [0002] With the maturity of mobile Internet technology, people can conveniently access the Internet through mobile devices at any time and place, and users' demand for information is becoming more and more personalized. Accurately predict the information needs of users at this moment, that is, real-time user intentions, It is an important aspect of personalization. [0003] The existing O2O (Online To Offline, Online To Offline, Online To Offline, Online To Offline) comprehensive search recommendation system generally adds the user's intention through historical behaviors such as clicking and placing an order as a feature into the learning ranking model. The business to be recommended to the user is obtained through traini...

AI Technical Summary

Problems solved by technology

However, in the absence of query, it does not reflect the user's instantaneous tendency very well, for example, the user's intention or tendency when he first launched the search system

At the same time, the use of multiple systems for user historical behavior characteristics requires multiple accesses, which increases the complexity of model calculations and increases the probability of system errors

Images