| dc.description.abstract | In the modern day, news suggestions have emerged as the most widely used method
for locating pertinent news content following the COVID-19 epidemic. Nevertheless, current
approaches barely uncover underlying connections between movies and are ignorant of such
vast amounts of external knowledge. Their ability to classify users based on which genres,
tags, or types of movies they are passionate about is limited. Consequently, in order to create
easier circumstances for users to obtain the context they want and to enhance the reading
experience, the recommendation system needs to extract individual features.
In this report, we offer a knowledge-aware CNN-improved global knowledge-aware
and local attention-aware recommendation system. The global knowledge-aware module
extracts the knowledge graph from news titles by means of each word, along with related
contexts and matching entities. The local attention-aware module examines how the current
news relates to other historical news stories.
Consequently, our methodology is able to accurately anticipate whether the user
would click on the candidate news or not based on the two modules mentioned above. The
MIND dataset serves as the training set. Regarding news recommendation, the global
knowledge-aware and local attention-aware framework has demonstrated impressive results
that surpass the state-of-the-art techniques in AUC, MRR, nDCG@5, and nDCG@10
assessment measures. | en_US |
