Semantic link based recommendation method and device for heterogeneous resource of TV platform

Abstract

The invention discloses semantic link based recommendation method and device for a heterogeneous resource of a TV platform. The method comprises the steps of extracting text information of all media resources from a background media resource library; extracting feature word candidates of the media resources on the basis of the text information of each media resource; calculating the weight of the feature word candidates; filtering the feature word candidates on the basis of the weight to obtain the feature words; producing a feature word weight matrix T of the background media resource library; calculating the cluster similarity of each media resource in the background media resource library and the current media resource by the cluster method on the basis of the feature word weight matrix T in case that the current media resource viewed by a user is the media resource in the background media resource library; and selecting L media resources with the maximum cluster similarity to produce a media resource recommendation list.

Description

technical field [0001] The invention relates to the field of multimedia technology, in particular to a method and device for recommending heterogeneous resources based on semantic links on a television platform. Background technique [0002] When a user watches a TV program on a TV platform, he is usually interested in some information about the current program, and wants to further watch other media resources related to the current program. Aiming at this psychology of users, there are currently some recommendation methods between media resources. Generally, according to the current resources watched by users, keywords of current resources are obtained to represent user characteristics, and the obtained keywords are used as vectors representing user characteristics. Recommend resources with high similarity to current resources to users. [0003] However, the current recommendation methods among these media resources have many disadvantages, for example: most of them are re...

AI Technical Summary

Problems solved by technology

[0003] However, the current recommendation methods among these media resources have many disadvantages, for example: most of them are recommendations among similar resources, and there are few recommendation applications among heterogeneous resources; most of the few heterogeneous resource recommendations are one-way recommendations, that is, From one resource to another, such as video source recommendation methods associated with TV programs, product recommendation methods associated with TV programs, etc., there are few methods for mutual recommendation between multiple resources; resource recommendation methods play an important role Some words are recognizable, some cannot be recognized and require manual construction, which is cumbersome to operate; limited to word form information, lacking semantic information; relying on manual annotation, lack of utilization of user feedback, and its recommendation results are not very good for users ideal

Images