Suggesting Accurate and Faithful Locale

Abstract

In recent years, the boundaries between Online Business and Online Communication have become increasingly blurred. Many Online Business websites support the mechanism of social login where users can sign on the websites using their social network identities such as their Face book or Twitter accounts. Users can also post their newly purchased products on micro blogs with links to the Online Business product web pages. In this paper we propose a novel solution for cross-site Problem In Crucial product Propose, which aims to recommend products from Online Business websites to users at Online Communication sites in “Problem In Crucial” situations, a problem which has rarely been explored before. A major challenge is how to leverage grasp extracted from Online Communication sites for cross-site Problem In Crucial product Propose. We propose to use the linked users across Online Communication sites and Online Business websites (users who have Online Communication accounts and have made purchases on Online Business websites) as a bridge to map users’ Online Communication features to another feature representation for product Propose. In specific, we propose learning both users’ and products’ feature representations (called user implants and product implants, respectively) from data collected from Online Business websites using recurrent neural networks and then apply a modified gradient boosting trees method to transform users’ Online Communication features into user implants. We then develop a feature-based matrix factorization approach which can leverage the learnt user implants for Problem In Crucial product Propose. Experimental results on a large dataset constructed from the largest Chinese micro blogging service SINA WEIBO and the largest Chinese B2C Online Business website JINGDONG have shown the effectiveness of our proposed Software Set.