Transaction preference discrimination model adaptive method

Abstract

The invention discloses a transaction preference discrimination model adaptive method, and relates to the technical field of user transaction preference discrimination. The method comprises eight sequential steps: establishing a transaction graph database, preprocessing data, extracting features, establishing a transaction behavior analysis model, establishing a transaction content analysis model,synthesizing the transaction behavior analysis model and the transaction content analysis model to carry out transaction preference judgment, performing verifying retrieval based on a transaction preference result, and dynamically optimizing a transaction preference discriminating model based on the verifying result. According to the method, the transaction behavior, the transaction content, thetransaction associated preference and the intelligent verification feedback are combined to judge the transaction preference; and the transaction preference discrimination model is dynamically optimized by learning a verifying retrieval method and the result feedback of professional personnel, so that the problem of hysteresis between the transaction preference discrimination model and transactionbehavior changes is solved, the robustness of the model and the accuracy of transaction preference recommendation are improved, and the method has important significance for discovering user trade preferences of massive trade data.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to discrimination of user transaction preferences, and in particular to a method for self-adaptation of a transaction preference discrimination model. Background technique [0002] The rapid development of the Internet has given birth to various electronic trading platforms, such as online trading platforms such as e-commerce, group buying, and food delivery. Changes have taken place. Electronic payment methods not only break down the barriers of offline and online transactions, but also greatly enhance the convenience of transactions and speed up the flow of payment transactions. [0003] In the application of various online trading platforms, in order to better capture users' trading preferences and trading habits, and realize precise marketing or product recommendations based on user characteristics, many applications of recommendation systems have emerged. For e...

AI Technical Summary

Problems solved by technology

[0004] However, the above-mentioned recommendation algorithm only makes recommendations based on the user's historical transactions, but does not really identify the user's transaction preferences and habits. After the user buys an item for others or makes an occasional transaction, the user will still recommend such items to the user, resulting in Recommendation accuracy is not high

When the number of users or the number of items are massive, the time cost of the above recommendation scheme is very high, and the accuracy of the recommendation is poor

At the same time, the above algorithm ignores the impact of changes in user behavior

Images