基于深度学习的文本分类系统关键技术研究与模型验证.PDFVIP

专栏：人工智能技术与应用 基于深度学习的文本分类系统关键技术研究与模型验证 汪少敏，杨迪，任华 （中国电信股份有限公司上海研究院，上海 200122 ） 摘 要：大数据时代，文本分类是文本数据挖掘和文本价值探索领域的重要工作。传统的文本分类系统存在 特征提取能力弱、分类准确率不高的问题。相对于传统的文本分类技术，深度学习技术具有准确率高、特征 提取有效等诸多优势，有必要将深度学习技术引入文本分类系统，以解决传统文本分类系统存在的问题。在 分析传统文本分类系统的基础上，提出了基于深度学习的文本分类系统的体系架构和关键技术，同时对传统 分类模型、TextCNN、CNN+LSTM 多种分类模型进行了验证比对。 关键词：深度学习；文本分类；分类模型；神经网络 中图分类号：TP393 文献标识码：A doi: 10.11959/j.issn.1000−0801.2018301 Key technology research and model validation of text classification system based on deep learning WANG Shaomin, YANG Di, REN Hua Shanghai Research Institute of China Telecom Co., Ltd., Shanghai 200122, China Abstract: Text classification is very important to text data mining and value exploration. The traditional text classifi- cation system has problems of weak feature extraction ability and low classification accuracy. Compared with the tra- ditional text classification technology, deep learning technology has many advantages such as high accuracy and ef- fective feature extraction. Therefore, it is necessary to apply deep learning technology to the text classification system to solve the problems of the traditional text classification system. The traditional text classification system was ana- lyzed, and the architecture and key technologies of text classification system based on deep learning were proposed. Finally, several classification models were verified and compared, including the traditional classification model, TextCNN and CNN+LSTM. Key words: deep learning, text classification, classification model, neural network 理由于其庞大的数据源、相对成熟的处理技术和 1 引言 广泛的应用需求，备受关注。文本处理中最典型 目前，人工智能领域能处理的三大类信息格