News Article Name Disambiguation Model Based on Reinforcement Learning
Yi Ding · 2021
With the advent of information media society, people are facing classification problem that whether two news articles describe the same event. This helps readers identify fake news and better accept social media information. An important point is to judge whether the names in the article are the same person by the name disambiguation method. In this paper, based on the disambiguation model of reinforcement learning, the author proposes an improved solution for news articles name disambiguation problem. Solution contains entity recognition algorithms based on news article, entity disambiguation model with paragraph training method and self-adjusting method based on reinforcement learning. The proposed solution is expected to have good availability and the HR@1 value of the system based on reinforcement learning is highly improved compared with the traditional method, which is indirectly proved by experiment.