A product collocation recommendation method and device based on attributes and titles

Abstract

The invention discloses a method and device for recommending product collocation based on attributes and titles. The method obtains collocation scores among commodities by acquiring user order information within a period of time, and can obtain products with higher collocation scores according to the conditions of collocation scores. Candidate product sets, and by introducing heterogeneous co-occurrence frequency and heterogeneous co-occurrence inverse frequency to reduce the impact of popular products on the collocation of products, so as to obtain a more accurate candidate product set; The vector of the title, and map the two types of vectors into the vector space, learn the collocation degree between two commodities, establish a commodity collocation model, and then obtain the collocation rate between commodities more accurately based on commodity attributes and titles; and then according to the user The purchased or inquired commodity information can match a candidate commodity with the highest matching degree from the candidate commodity set of the commodity. Based on this method, the device aims at accurately recommending matching commodities.

Description

technical field [0001] The invention relates to the field of collocation recommendation, in particular to a product collocation recommendation method and device based on attributes and titles. Background technique [0002] With the development of e-commerce, online shopping has gradually become the mainstream way of shopping. In shopping, merchants usually hope that users can recommend matching products after purchasing or paying attention to a certain product, or when users ask questions such as "what kind of pants should I wear with this T-shirt I bought", they can quickly and Accurate answers are given to ultimately improve user shopping experience and promote product sales. Such application scenarios are more common in clothing shopping. Although the matching and combination configuration of clothes can solve the above problems by human beings, it is undoubtedly a time-consuming and difficult task. At the same time, it requires the configuration personnel to have profes...

AI Technical Summary

Problems solved by technology

However, the number of words in the title is limited, and the characteristics of the product are obtained simply through the title, and many important information of the product will be missed.

On the other hand, training these models requires a large collocation dataset labeled by experts, which is also very time-consuming

Images