The paper "DKN: Deep Knowledge-Aware Network for News Recommendation" addresses the challenge of personalized news recommendation in the context of information overload. It proposes a deep knowledge-aware network (DKN) that integrates knowledge graph representation into news recommendation to enhance the discovery of latent knowledge-level connections among news items. DKN is a content-based deep recommendation framework designed for click-through rate (CTR) prediction. The key component of DKN is a multi-channel and word-entity-aligned knowledge-aware convolutional neural network (KCNN), which fuses semantic-level and knowledge-level representations of news. KCNN treats words and entities as multiple channels and aligns their representations during convolution. Additionally, an attention module is designed to dynamically aggregate a user's history with respect to current candidate news, addressing the diversity of user interests. Extensive experiments on a real online news platform demonstrate that DKN outperforms state-of-the-art deep recommendation models, achieving substantial gains in F1 and AUC metrics. The results also validate the effectiveness of using knowledge and the attention module in DKN.The paper "DKN: Deep Knowledge-Aware Network for News Recommendation" addresses the challenge of personalized news recommendation in the context of information overload. It proposes a deep knowledge-aware network (DKN) that integrates knowledge graph representation into news recommendation to enhance the discovery of latent knowledge-level connections among news items. DKN is a content-based deep recommendation framework designed for click-through rate (CTR) prediction. The key component of DKN is a multi-channel and word-entity-aligned knowledge-aware convolutional neural network (KCNN), which fuses semantic-level and knowledge-level representations of news. KCNN treats words and entities as multiple channels and aligns their representations during convolution. Additionally, an attention module is designed to dynamically aggregate a user's history with respect to current candidate news, addressing the diversity of user interests. Extensive experiments on a real online news platform demonstrate that DKN outperforms state-of-the-art deep recommendation models, achieving substantial gains in F1 and AUC metrics. The results also validate the effectiveness of using knowledge and the attention module in DKN.