TagTag: A Novel Framework for Service Tags Recommendation and Missing Tag Prediction

Wentao Chen, Mingyi Liu, Zhiying Tu, Zhongjie Wang

Abstract: Currently, service tag recommendation plays an important role in the study of services. As a result, there have been many service tag recommendation studies that have achieved significant achievements. However, existing studies mainly have two problems: they only recommend one tag and cannot determine whether new tags are needed. To help solve the above problems, we propose a novel graph neural framework named TagTag to make multi-tag recommendations and missing tag prediction, which relies on the idea of tag collaboration graph. We conduct experiments on the real-world dataset from ProgrammableWeb, and the results show that TagTag performs better than existing studies. The code used in this paper is fully accessible at https://github.com/HIT-ICES/TagTag.