On the current state of deep learning for news recommendation

The exponential outbreak of news articles makes it troublesome for the readers to find, select and read the most relevant ones and alleviate the resulting information and cognitive overload problems. In recent years, various deep learning (DL) models were developed to recommend personalized quality news articles and support readers. Yet, no survey paper highlights the strengths, weaknesses, and trends of news recommendation models employing DL methods. This survey fills this gap in the literature by identifying the current state of DL-based news recommendation methods. Specifically, it covers (1) the classification of DL-based news recommendation models, (2) their performance comparison, and (3) the essential issues faced by these models. It discusses the most commonly used datasets, evaluation methods, and implications for researchers working in this area.