Potential user recommendation method based on service multi-granularity attributes

Abstract

The invention provides a potential user recommendation method based on service multi-granularity attributes. The potential user recommendation method comprises the following steps that S1, according to a service common type class cluster generated by clustering service containing type tags, carrying out user scoring prediction on service coarse-grained attributes; S2, calculating the similarity between services based on the service inclusion type according to the jaccard coefficient, and carrying out the scoring prediction of the fine-grained attribute of the service by the user according to the neighbor service; S3, performing weighted summation on the prediction scores of the coarse-grained attributes and the fine-grained attributes of the services, implementing prediction on the multi-granularity attributes of the services by a user, sorting the prediction scores, and selecting top-k potential users with higher ratings for recommendations. The method has the advantages that recommendation research of potential users is carried out from the perspective of service oriented. The coarse-grained attribute and the fine-grained attribute of the service are considered from the class cluster of the common type of service and the type of the service, so that the cold start problem is effectively solved, and recommendation accuracy is improved.

Description

technical field [0001] The invention belongs to the technical field of service computing, and in particular relates to a potential user recommendation method based on multi-granularity attributes of services. Background technique [0002] With the in-depth development of the Web2.0 era, people are already in an Internet environment that pays more attention to interaction. People no longer simply stand from the perspective of information acquirers, but have become more involved, influencing and changing Internet information with their own habits. In addition, information aggregation leads to the continuous accumulation of Internet information, making it increasingly difficult for users to select information for individual needs. [0003] The recommendation system is a technology based on data mining. According to the analysis of massive data, it can recommend personalized decisions and relevant information to users. For example, Dangdang.com can provide readers with various...

AI Technical Summary

Problems solved by technology

However, the existing recommendation methods still have the following problems: 1), the premise of using collaborative filtering method is that a service is used by at least one user, so as to ensure that the service may be recommended, but a new service is often not used by other users Use and scoring, that is, the cold start problem faced by new services; 2), content-based recommendation methods are easily affected by content analysis technology, and it is difficult to effectively identify service characteristics, resulting in inaccurate recommendation results; 3), existing recommendations Most of the technologies recommend services to users, but do not recommend potential users from the perspective of services

Images