计算机工程与应用2019,Vol.55Issue(18):111-115,5.DOI:10.3778/j.issn.1002-8331.1903-0232
基于CRF的藏文地名识别技术研究
Research on Tibetan location Name Recognition Technology Under CRF
摘要
Abstract
Tibetan location name recognition is a problem that must be solved in Tibetan named entity recognition. By ana-lyzing the characteristics and recognition difficulties of Tibetan location names, this paper expounds that the characteris-tics of syllables, trigger words, location name follow-up words and case auxiliary words of Tibetan location names are applicable to location name recognition based on CRF model. Through experiments, the effectiveness of the six character-istics of this paper on Tibetan location name recognition is verified. The experimental results show that the accuracy rate, recall rate and F value of Tibetan location name recognition by this method reach 96.12%, 81.92% and 88.45%, respec-tively. Compared with the existing systems, the experimental results have achieved better results.关键词
CRF模型/藏文地名/地名识别Key words
CRF model/Tibetan location/location name recognition分类
信息技术与安全科学引用本文复制引用
头旦才让,仁青东主,尼玛扎西..基于CRF的藏文地名识别技术研究[J].计算机工程与应用,2019,55(18):111-115,5.基金项目
国家重点研发计划重点专项(No.2017YFB1402200) (No.2017YFB1402200)
国家自然科学基金(No.61262051) (No.61262051)
青海省科技计划项目(No.2017-GX-146,No.2017-ZJ-767) (No.2017-GX-146,No.2017-ZJ-767)
西藏大学研究生"高水平人才培养计划"项目(No.2017-GSP-016). (No.2017-GSP-016)