摘要
基于自然语言理解的相似度计算仍是计算机语言处理技术尚需深入研究的内容。通过在"知网"知识表示的基础上,综合考虑深度和密度两方面的影响因素,利用一种较为成熟的改进的多因素语义相似度处理算法,基于全文检索匹配技术,设计并实现了一个限定领域内的在线答疑系统。实例运行结果表明,系统可靠性较高,且答疑效果较为明显,达到了预期目标。
The similarity calculation based on natural language understanding is still a research content of the computer language processing technology. Based on the knowledge representation of ＂HowNet＂, considering the both factors of depth and density, by using a more sophisticated multivariate semantic similarity algorithm, and with a full-text search matching technology, an online answer system in the limited field is designed and implemented. The experimental results show that, the system is reliable, the answer effect is more obvious, and the desired goal is achieved.
出处
《计算机时代》
2015年第5期10-12,共3页
Computer Era
关键词
自然语言理解
语义相似度
全文检索
在线答疑系统
natural language understanding
semantic similarity
full text retrival
on-line answer system
