ABSTRACT The emerging patent trading platforms help to ease information asymmetry and trust issues during transaction, but a proactive recommendation mechanism that intelligently helps patent buyers identify relevant patents is still absent in the literature. This study proposes a recommendation mechanism for patent trading empowered by heterogeneous information networks (HIN) that integrates various patent information such as patent trading, patent invention, patent citation, patent ontology, and patent contents. Further, the meta-path-based similarity measure (i.e., AvgSim) is employed to calculate relevance and identify the different motivations of potential buyers in buying patents. We conducted two experiments to examine the performance of a proposed mechanism. An offline experiment on Public PatentsView database and Patent Assignment database show that the HIN-empowered recommendation outperforms baseline methods. We also implemented the proposed mechanism on a real-world trading platform (www.InnoCity.com). The recommendation function achieves satisfying results by tracking users’ feedback, which further validates the usability of HIN-empowered recommendation in a patent trading context.

中文翻译：

---

异构信息网络赋能专利交易推荐机制

摘要 新兴的专利交易平台有助于缓解交易过程中的信息不对称和信任问题，但文献中仍然缺乏一种能够智能帮助专利购买者识别相关专利的主动推荐机制。本研究提出了一种由异构信息网络（HIN）赋能的专利交易推荐机制，该机制集成了专利交易、专利发明、专利引用、专利本体和专利内容等各种专利信息。此外，采用基于元路径的相似性度量（即 AvgSim）来计算相关性并识别潜在购买者购买专利的不同动机。我们进行了两个实验来检查所提出机制的性能。Public PatentsView 数据库和专利转让数据库的离线实验表明，HIN 授权的推荐优于基线方法。我们还在真实世界的交易平台 (www.InnoCity.com) 上实施了提议的机制。推荐功能通过跟踪用户的反馈获得了令人满意的结果，这进一步验证了 HIN 授权推荐在专利交易环境中的可用性。