Interdisciplinary recommending method and system

Abstract

The invention discloses an interdisciplinary recommending method and system. The method comprises the steps of determining behavior preference data of a user in multiple fields, and then utilizing a recommendation algorithm and the behavior preference data to generate a recommendation result for the user. This shows that the behavior preference data is obtained from multiple fields instead of being obtained from a single field, in this way, data sources in an information recommending process can be obviously drastically enriched, and the data acquisition difficulty is thus lowered; meanwhile, the behavior preference data of the user in multiple fields can more comprehensively embody the individual interest of the user, and thus the follow-up recommendation result can be sharply improved.

Description

technical field [0001] The invention relates to the technical field of information recommendation, in particular to a cross-domain recommendation method and system. Background technique [0002] With the development of social networks, ubiquitous computing, and e-commerce, the territory of the Internet has been greatly expanded, and we have entered the era of "information trek". In the early days, search engines were able to solve the problem of finding information for us. However, with the advent of the era of big data, how do users find the information they are interested in among the massive amounts of information? The recommendation system was born to allow users to find information that they may be interested in. [0003] Existing recommendation systems are mainly limited to a single field, and they recommend content of interest to users based on their behavior characteristics in a certain field. However, single-domain recommendation has obvious flaws. First of all, ...

AI Technical Summary

Problems solved by technology

However, single-domain recommendation has obvious flaws

First of all, the current algorithms of single-domain recommendation systems basically use user-based collaborative filtering algorithms. There are two major difficulties in this algorithm, one is the cold start problem; the other is the sparsity problem of the matrix

Secondly, from a psychological point of view, people's interests and hobbies are diversified, so a single field of interest cannot reflect a complete personal interest

That is to say, in the existing information recommendation process, there are problems such as relatively single data source and difficult data collection, and the recommendation effect is not satisfactory

Images