基于结构化语料库的最优概念语义模板自检索  

Optimal Conceptual Semantic Template Self Retrieval Based on Structured Corpus

作　　者：尚雪茹 陈翰 SHANG Xue-ru;CHEN Han(Sichuan Tianyi College,Deyang Sichuan 618200,China)

机构地区：[1]四川天一学院,四川德阳618200

出　　处：《计算机仿真》2025年第1期550-554,共5页Computer Simulation

基　　金：成都市应用心理学研究会科研规划项目(CYXYJH2304)。

摘　　要：语义模板语料库具有大规模性,直接从语料库中搜索最优语义模板十分困难,降低了后续检索效率。为了提高语义模板检索效果,提出基于结构化语料库的最优概念语义模板自检索。首先,对语料库中的各类语义模板展开聚类处理,利用语料库结构化目标函数,建立结构化语料库,缩短语义模板的检索时间;其次,结合多尺度特征融合与注意力机制建立语义特征提取网络,提取语料库语义特征,为后续语义检索提供依据;最后,计算用户翻译文本与结构化语料库中语义模板之间的概念相似度和属性相似度,在此基础上计算语义相似度,选取语义相似度最高的语义模板作为最优概念语义模板检索结果。实验结果表明,所提方法的检索效率高、精度高。The semantic template corpus is large-scale,and it is very difficult to search the optimal semantic template directly from the corpus,which reduces the efficiency of subsequent retrieval.In order to improve the retrieval effect of semantic templates,we propose the self-retrieval of optimal conceptual semantic templates based on structured corpus.Firstly,the various types of semantic templates in the corpus are clustered,and the structured objective function of the corpus is utilized to establish a structured corpus to shorten the retrieval time of the semantic templates;Secondly,a semantic feature extraction network is established by combining the multi-scale feature fusion and the attention mechanism,and semantic features are extracted from the corpus to provide a basis for the subsequent semantic retrieval;Finally,we calculate the concept similarity and attribute similarity between user translated text and semantic templates in the structured corpus,on the basis of which we calculate the semantic similarity and select the semantic template with the highest semantic similarity as the optimal conceptual semantic template retrieval result.The experimental results show that the proposed method has high retrieval efficiency and high accuracy.

关 键 词：结构化语料库 注意力机制 语义模板检索 特征融合 语义相似度 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]