融合LSTM的自然语言转结构化查询语句算法的研究与设计  

Research and Design of Natural Language to Structured Query Language with LSTM

作　　者：孙红[1] 黄瓯严 SUN Hong;HUANG Ou-yan(School of Optical-Electrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海200093

出　　处：《小型微型计算机系统》2023年第1期63-67,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61472256,61170277,61703277)资助;上海市教委科研创新重点项目(12zz137)资助;沪江基金项目(C14002)资助。

摘　　要：自然语言转结构化查询语句(Natural Language to SQL,NL2SQL)是信息领域一个重要课题.目前前沿的NL2SQL工作都是针对英文数据集,而处理英文数据的方法直接应用到中文上往往难以取得很好的效果.本文首先对传统的SQLNet模型进行了改进,在其中融入了预训练模型,增强了其提取特征的能力;之后又分别对分类模型和条件值模型进行了改进:在分类模型中增加了LSTM进一步捕捉特征,在条件值模型中使用正则表达式等手段对特殊的条件子句进行了预处理.实验表明,本文对分类模型和条件值模型所做的改进都能有效提升模型的表达效果.The conversion of natural language to SQL language(Natural Language to SQL,NL2 SQL)is an important topic in the field of information.The current cutting-edge NL2 SQL work is all aimed at English data sets,and it is often difficult to achieve good results when the methods of processing English data are directly applied to Chinese.This article first improves the traditional SQLNet model,incorporates a pre-training model,and enhances its ability to extract features;later,the classification model and the conditional value model are improved separately:LSTM is added to the classification model to further capture Features,using regular expressions and other means in the condition value model to preprocess the special condition clauses.Experiments show that the improvements made in this article to the classification model and the conditional value model can effectively improve the expression effect of the model.

关 键 词：自然语言处理 自然语言转结构化查询语句 长短时记忆网络 预训练模型 查询化结构语言网络模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]