BiMuF: a bi-directional recommender system with multi-semantic filter for online recruitment

Pei-Yuan Lai, Zhe-Rui Yang, Qing-Yun Dai, De-Zhang Liao, Chang-Dong Wang

Abstract: Most existing recommendation research has been concentrated on unidirectional recommendation, i.e., only recommending items to users. However, the platform needs to achieve bi-directional recommendation in many real-world scenarios. For example, in an online recruitment scenario, the recommender system not only needs to recommend positions to candidates, but also recommend candidates to enterprises. In this paper, we develop a new bi-directional recommendation model for online recruitment termed (BiMuF) Bi-directional recommendation with Multi-semantic Filter. In BiMuF, an encoder component is utilized to learn the text embeddings, a multi-semantic filter component is designed to capture important graph representation, and a graph learning component is designed to learn the graph embeddings. In addition, a multi-task learning framework is designed to achieve bi-directional recommendations. In the multi-task learning framework, we share text embeddings and graph embeddings to alleviate the problems of data sparsity, data asymmetry, and feature generalization in online recruitment. Moreover, we conduct three new datasets based on a technology supply and demand docking platform namely South China Technology Commercialization Center, and three new datasets in job recruitment scenario namely Computer Technology-Related Job Recruitment. Extensive experiments in the real-world tasks show that BiMuF outperforms the state-of-the-art methods, verifying the effectiveness of the designs of our model. The code and the latter datasets are available at https://github.com/allminerlab.