基于句子向量表示和模糊C均值的电子政务文档自动摘要技术

Automatic Summarization of e-Government Documents Based on Sentence Vector Representation and Fuzzy C-Means

【目的】随着“互联网+电子政务”的发展,国家越来越重视我国电子信息化建设,对于政府相关决策者、管理者、信息化工作者及研究人员来说,迫切需要一种方式可以快速有效地获取众多的电子政务资讯来指导信息化评估和决策。本文旨在研究一种适合电子政务文档的自动摘要算法。【方法】本文针对电子政务资讯文本的特点提出了一种融合Doc2Vec句子向量表示方法和模糊均值聚类方法的算法并应用在电子政务资讯文档的自动摘要生成中,不仅考虑句子之间的相关度,而且针对文章的特点对于每个句子赋予一定的权重来表示他作为摘要句子的重要性。【结果】实验表明,相较于目前常用的k-means算法结果和复杂的深度学习算法结果,该算法在电子政务资讯文档的自动生成取得了比较好的结果。【结论】研究自动摘要技术并在电子政务领域应用是一项很有价值的工作。

[Objective]With the development of"Internet+E-Government",more and more attention has been paid to the construction of electronic information technology in China.For government decisionmakers,managers,information workers and researchers,there is an urgent need to quickly and effectively obtain plenty of E-Government information to guide information evaluation and decisionmaking.This paper studies an automatic summarization algorithm for e-government documents.[Methods]According to the characteristics of e-government information text,this paper proposes an algorithm that uses Doc2Vec sentence vector representation and fuzzy c-means to automatically generate the summary of e-government information documents.It not only considers the correlation between sentences,but also gives weight to each sentence to express its importance as a summary sentence according to the characteristics of the article.[Results]Experiments show that,compared with the commonly used k-means algorithm and complex deep learning algorithms,this algorithm achieves better results in automatic generation of e-government information documents.[Conclusions]The proposed algorithm is effective for automatic document digest in the field of e-government.