The research and application of the Chinese sports information extraction

By combining the model framework and rules of the unified conditional random fields, a new approach is proposed in this paper. We investigate the two key technologies of Information Extraction (IE), known as Named Entity Recognition (NER) and automatic entity Relation Extraction (RE), and explore several new features of our model on the “sports game” scene. The proposed method is implemented on some data collected from sohu.com and sina.com. Experimental results show that by combining the above new elements, our approach brings some improvements to IE, where the recall, precision and the F-measure are 95.70%, 93.00% and 94.33% respectively.