基于知识图谱的金融新闻个性化推荐算法

个性化新闻资讯推荐能够有效地捕捉用户兴趣，提供高质量推荐服务的能力，因而吸引了大量高黏性用户，而知识图谱则以“实体-关系-实体”的形式表示事物间的关系，通过知识图谱中实体间的关系学习到更丰富的特征及语义信息。为更好地实现金融领域新闻的个性化推荐，提出一种基于知识图谱的个性化推荐算法KHA-CNN。结合金融业知识图谱，采用基于知识的卷积神经网络和层次注意力机制得到新闻文本的特征表示，并学习用户复杂行为数据特征。在真实数据集上的实验结果表明，与Random Forest、DKN、ATRank-like算法相比，KHA-CNN算法的F1和AUC指标分别提高了2.6个和1.5个百分点。

Personalized Recommendation Algorithm for Financial News Based on Knowledge Graph

Personalized news information recommendation can attract a large number of highly sticky users because of its ability to effectively capture user interests and provide high-quality recommendation services.Knowledge graph represents the relationships between things in the entity-relation-entity form, which enables the learning of richer features and semantic information.To increase the quality of personalized recommendation of news in the financial field, this paper proposes a personalized recommendation algorithm, KHA-CNN, based on knowledge graph.Combined with the knowledge graph in the financial industry, a knowledge-based convolutional neural network and the hierarchical attention mechanism are used to obtain the feature representation of news texts, and to learn the features of the complex behavior data of users.Experimental results on real data sets show that compared with Random Forest, DKN, and ATRank-like algorithms, the KHA-CNN algorithm increases the F1 score by 2.6 percentange points, and the AUC indicator by 1.5 percentange points.