Personalized news recommendation methods can reveal users’ individualized demands for news information and deliver tailored content. Comprehensively and accurately modeling news and user interests is crucial for these recommendation methods. Several existing methods accomplish news modeling by extracting and fusing multiple news attributes, while user-interest modeling relies on multiple news representations. However, two pervasive issues often hinder recommendation performance: frequently overlooking potential limitations within attribute information such as topics, titles, and abstracts, and neglecting to explore user interests from diverse news attribute perspectives. We propose a Structured Topic-enhanced news Recommendation method with multi-view learning (STRec). This method initially constructs a topic thesaurus based on topics and their contexts to enhance topic representation. Subsequently, it combines titles and abstracts to form a content description and completes feature extraction based on three aspects of attribute information: structured topics, contents, and entities. Finally, the multi-view attribute features are fused to form representations of news and users, which are then used for personalized recommendations. Comprehensive experiments on real-world news datasets reveal that STRec can significantly improve the recommendation performance and that structured topic features and multi-view learning strengthen the representation of news and user interests.