融合词性与语义相关性的图书馆智能咨询系统问句相似性计算方法研究
Research on the Calculation Method of Question Similarity Base on the Integration of Parts-of-Speech and Semantic Correlation in Library Intelligent Consulting System
周朝阳 1贺艳菊 2夏岭梅 2闵昭浩2
作者信息
- 1. 湖北大学图书馆 湖北武汉 430062;湖北大学历史文化学院 湖北武汉 430062
- 2. 湖北大学图书馆 湖北武汉 430062
- 折叠
摘要
[目的/意义]提高问句相似性计算的准确性,对改善图书馆智能咨询的实际效果具有重要作用.[方法/过程]引入特征词在问句之间的语义相关性指标,提出了融合词性与语义相关性的问句相似性计算方法.首先构建咨询领域专业词典以增强问句处理的业务针对性,然后提出特征词词性权重和语义相关性权重度量方法,并融合两者度量词语在问句中的作用以构建问句语义表征向量,最后采用余弦函数计算问句相似度.[结果/结论]实验表明综合词性与语义相关性特征能有效识别问句中的核心特征词,本文提出的问句相似性计算方法比传统方法具有更高的ACC@1和ACC@3准确率.
Abstract
[Purpose/significance]Improving the accuracy of question similarity calculation plays an important role in improving the practical effectiveness of library intelligent consulting.[Method/process]Introducing the semantic correlation index of feature words between questions,this paper proposes a question similarity calculation method that combines part-of-speech and semantic correlation.Firstly,the paper constructs a professional dictionary in the consuulting field to enhance the business pertinence of question processing.Then,it proposes a feature word part-of-speech weight and semantic correlation weight measurement method,and combines the above two to measure the role of words in the question to construct a question semantic representation vector.Finally,it uses the cosine func-tion to calculate the quetion similarity.[Result/conclusion]Experiments show that the combination of part-of-speech and semantic cor-relation features can effectively identify the core feature words in questions.The proposed method has a higher accuracy rate of ACC@1 and ACC@3 than the traditional method.
关键词
词性/语义相关性/相似度计算/咨询服务/智能咨询Key words
part-of-speech/semantic correlation/similarity calculation/consulting service/intelligent consulting引用本文复制引用
基金项目
湖北省高校图工委科研基金研究项目(2021-ZD-04)
出版年
2024
NETL
NSTL
维普
万方数据