大数据背景下公平竞争审查文本的排查研究

Research on the Investigation of Fair Competition Review Text in the Context of Big Data

在现阶段,各地市场监管局开展公平竞争审查的方式主要是依据政府监管人员对审查公文进行逐一排查,客观上受监管人员能力和主观性判断的限制,且人工排查条件下效率较低.针对这种情况,提出了建立公平竞争审查的关键词词典,基于关键词过滤的方法进行排查和应用深度学习中的文本卷积神经网络进行排查的两种方法,帮助审查人员提高排查效率和扩大实务中的排查范围.将广东省市场监督管理局提供的数据和爬取广东省厅级单位的全部公开文件的数据导入到MySQL数据库,从这些数据的标题中人工筛选出违反了公平竞争审查标准的关键词和没有违反的关键词,并基于编写的SQL程序通过关键词过滤的方法筛选文本,随后再利用文本卷积神经网络对数据的主体内容进行自动分类.实验表明两种方法都有助于提升文本排查的效率,但基于卷积神经网络的方法效果更好以及更加符合当前的实际需求.

The main way for local market supervision bureaus to conduct fair competition review is that government regulators conduct one-by-one inspections of the review documents.This method is objectively restricted by the ability and subjective judgment of the regulators,and is less efficient under manual investigation conditions.In response to this situation,this paper proposes two methods of establishing a keyword dictionary for fair competition review,searching based on keyword filtering,and applying deep learning text convolutional neural networks to help examiners improve the efficiency of the investigation and expand the scope of investigation in practice.Two parts of the data provided by the Guangdong Provincial Market Supervision Administration are imported.All the public documents are crawled from the Guangdong Provincial Department-level units into the MySQL database.Keywords are manually selected that violate the fair competition review standards from the titles of these data and keywords that do not violate the review standards.The text is filtered,based on the written SQL program through keyword filtering;finally.Text convolutional neural network is used to automatically classify the main content of the data.Experiments show that these two methods are helpful to improve the efficiency of text screening,and the method based on convolutional neural network is more effective and more in line with current actual needs.