基于知识图谱的在线个性化商品推荐技术研究

With the rapid development of artificial intelligence (AI) in the age of big data, personalized item recommender systems (PIRS) have been playing a more and more important role in various intelligent electronic commerce platforms. However, there still exists many challenges in the research field of PIRS. Among these challenges, two key problems are: (1) consistent representation and fusion of multi-source heterogeneous information, and (2) design of online recommendation strategies and algorithms. In fact, they are two critical factors which influence the effectiveness and efficiency of PIRS, and thus have significant research values and broad application prospects. In this project, we intend to explore knowledge graph (KG) techniques both systematically and theoretically considering the consistent representation and fusion of multi-source heterogeneous information, and will also propose new ideas and solutions to key technologies in online PIRS with the integration of multi-domain’s KGs. Specifically, we will focus on the following five aspects, including “design of KG”, “knowledge association mining”, “online strategies and algorithms of PIRS based on KG”, “online strategies and algorithms of PIRS based on integrated KGs”, and “the evaluation of online PIRS”. The proposed methods are expected to enrich the theoretical achievement in the field of PIRS, and to provide technical support for the realization and applications of PIRS.

随着大数据时代的到来和人工智能技术的迅猛发展，个性化商品推荐系统已成为智能化电子商务平台的重要组成部分。然而，个性化商品推荐系统仍面临着很多挑战，其中，亟待解决的两个核心问题是：多源异构信息的一致性表达和融合，以及在线推荐策略和算法的设计，它们作为影响推荐系统准确性和时效性的关键因素，具有重要的研究价值和广阔的应用前景。本项目拟结合知识图谱这一热点研究领域，对在线个性化商品推荐技术进行较为系统化、理论化的探索，在利用知识图谱技术对多源异构信息进行一致性表达和融合的基础上，为在线个性化商品推荐系统中的关键技术提供新的设计思路和解决方案。本项目拟重点研究“知识图谱设计”、“知识关联挖掘”、“基于知识图谱的在线个性化商品推荐策略和算法”、“基于集成知识图谱的在线个性化商品推荐策略和算法”以及“效用评估”等几个方面。所提出的新方法可望丰富个性化商品推荐系统领域中的理论成果,并提供新的技术支撑。

随着大数据时代的到来和人工智能技术的迅猛发展，个性化商品推荐系统已成为智能化电子商务平台的重要组成部分。然而，个性化商品推荐系统仍面临着很多挑战，其中，亟待解决的两个核心问题是：多源异构信息的一致性表达和融合，以及在线推荐策略和算法的设计，它们作为影响推荐系统准确性和时效性的关键因素，具有重要的研究价值和广阔的应用前景。本项目结合知识图谱这一热点研究领域，对在线个性化商品推荐技术进行较为系统化、理论化的探索，在利用知识图谱技术对多源异构信息进行一致性表达和融合的基础上，为在线个性化商品推荐系统中的关键技术提供新的设计思路和解决方案。本项目的主要研究内容包括：知识图谱设计、知识关联挖掘、基于知识图谱的在线个性化商品推荐策略和算法、基于集成知识图谱的在线个性化商品推荐策略和算法，以及效用评估等。本项目所提出的新方法为个性化商品推荐系统领域提供了一种新的技术支撑。