摘要
关键词作为学术文本中映射全文主题内容的词汇或术语,能够为知识精准检索和文本大规模计算提供重要的底层语义标签。当前学术文本中的关键词存在使用意图不明、语义功能模糊及上下文信息缺失等问题。为此,本文提出了一种基于有监督学习的神经网络方法,对关键词所承载的语义功能进行分类,实现对学术文本中研究问题和研究方法的识别。本文以计算机等领域为期10年的学术期刊论文为训练语料,利用BERT及LSTM方法构建分类模型,实验结果显示,本文所提出的方法较传统更优,其整体准确率、召回率和F1值分别达到0.83、0.87和0.85。
As vocabulary or terminology that maps the full-text subject matter content in academic texts,keywords can provide important underlying semantic labels for knowledge retrieval and large-scale text computation.At present,there are problems in the use of keywords in academic texts,such as unclear intention,fuzzy semantic function,and lack of context information.Therefore,a neural network method based on supervised learning is proposed to classify the semantic functions carried by keywords to facilitate the identification of research questions and research methods in academic texts.In this study,journal papers published during a period of 10 years in the field of computer science were used as the training corpus,and the classification model was constructed using BERT and LSTM models.The results show that the proposed method is better than the traditional method.Its overall accuracy,recall rate,and F1 value reached 0.83,0.87,and 0.85.
作者
陆伟
李鹏程
张国标
程齐凯
Lu Wei;Li Pengcheng;Zhang Guobiao;Cheng Qikai(School of Information Management,Wuhan University,Wuhan 430072;Institute for Information Retrieval and Knowledge Mining,Wuhan University,Wuhan 430072)
出处
《情报学报》
CSSCI
CSCD
北大核心
2020年第12期1320-1329,共10页
Journal of the China Society for Scientific and Technical Information
基金
国家社科基金重大项目“基于认知计算的学术论文评价理论与方法研究”(17ZDA292)。
关键词
学术文本
关键词
语义功能识别
深度学习
academic text
keywords
lexical function recognition
deep learning
