基于BIGRU的番茄病虫害问答系统问句分类研究  

Question Classification of Tomato Pests and Diseases Question Answering System Based on BIGRU

作　　者：赵明[1] 董翠翠 董乔雪[1] 陈瑛[1] ZHAO Ming DONG Cuicui DONG Qiaoxue CHEN Ying(College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, Chin)

机构地区：[1]中国农业大学信息与电气工程学院,北京100083

出　　处：《农业机械学报》2018年第5期271-276,共6页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金项目(61503386)

摘　　要：问句分类作为问答系统的关键模块,对系统检索效率具有决定性作用。为了对番茄病虫害智能问答系统用户问句进行高效分类,构建了基于word2vec和双向门控循环单元神经网络(Bi-directional gated recurrent unit,BIGRU)的番茄病虫害问句分类模型。针对问答系统对用户问句的语义信息有较高要求的特点,首先利用word2vec将句子中的词转换为具有语法、语义信息的词向量,利用训练得到的词向量和BIGRU神经网络进行问句分类模型的训练。实验选取了2 000个番茄病虫害相关的用户问句,主要分为番茄病害和番茄虫害两类。结果表明,采用BIGRU的番茄病虫害问句分类模型,其分类准确率、召回率和准确率与召回率的调和平均值F1分别高于卷积神经网络(CNN)、K最近邻等分类算法2~5个百分点。BIGRU模型结构简单,模型训练参数较少,模型训练速度快,符合问答系统对响应时间的要求。The notable feature of a question answering system is to understand the semantic information of the user＇s question. Question classification,as the key module of question answering system,plays a decisive role in the efficiency of system retrieval. In order to classify the user＇s questions,a classification model of tomato pests and diseases based on word2 vec and bi-directional gated recurrent unit（ BIGRU）was constructed. word2 vec was used to transform the words in the sentence into the word vector with semantic information. The word vector was used as the initial corpus. Two neural network methods and a machine learning method were adopted to train the classification model. Totally 2000 tomato pests and diseases related questions were selected,which were divided into two categories： tomato diseases and tomato pests. The results showed that the classification accuracy,recall rate and F1 value by using the BIGRU model were 2 ～ 5 percentage points higher than those by using convolutional ceural network（ CNN） and K-nearest neighbor（ KNN） classification algorithm. Further experimental results comparison indicated that the BIGRU model performed the best on tomato pest and diseases question classification.The BIGRU model was simple in structure,less in model training parameters,and fast in training speed.It met the response time requirements of question answering system.

关 键 词：番茄 病虫害 问答系统 问句分类 BIGRU 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程] S43[自动化与计算机技术—控制科学与工程]