Research on Sports News Oriented Chinese NER
Weiwen Wang · 2022
In the field of sports news, data analysis and named entity recognition(NER) around news content can provide a data basis for subsequent tasks such as building a sports knowledge database and recommendation system. Therefore, named entities in the field of sports news are of great value, but there are few studies in this field at present. This paper takes Chinese sports news NER as the primary research object, make an in-depth study on the entity characteristics, and designs the model structure of multi-feature fusion. As there are few data sets in this field, this paper also constructs a data set in the sports news field, and verifies the effectiveness of our model for Chinese named entity recognition. At the same time, it conducts comparative experiments with popular depth-learning methods. The main work of this paper is as follows: (1) A dataset for sports news Chinese NER is constructed, Includes 7,645 labeled samples and 14,258 entities in 6 categories (2) Aiming at the problems of low utilization rate of text features and insufficient semantic representation in the process of NER, this paper proposes a multi-feature fusion recognition model that integrates radical, character and word level features. The experimental results show that the recognition performance on our and other public datasets is better than the existing methods.