【摘要】 针对目前农业问答系统领域缺少语料库的问题,使用Python爬虫获取了5类共计28 608条问答句对文本数据,并对获取的文本数据进行了数据清洗、分词等预处理,构建了一个可用于问答系统研究的农业文本语料库.对获取问句进行分析,发现问句文本长度集中在30个字符以内,列出了各种文本的主题分布情况.采用Text CNN模型对问句进行分类,最好的宏平均F1值为88.762,结果证明本文所建语料的可用性,带标注的语料库构建对农业问答领域的研究具有重要意义.更多还原

【Abstract】 In view of the lack of corpus in the field of agricultural question answering system,Python crawler was used to obtain a total of 28 608 question answering sentences in five categories for text data,and the obtained text data was preprocessed by data cleaning and word segmentation,and an agricultural text corpus for question answering system research was constructed.By analyzing the obtained questions,it is found that the length of the question text is concentrated within 30 characters,and the topic distribution of various texts is listed.Using TextCNN model to classify questions,the best macro-average F1 value is 88.762,which proves the availability of the corpus constructed in this paper.The construction of annotated corpus is of great significance to the research in the field of agricultural question answering.更多还原