近代报刊新闻报道的时间线抽取——基于《东方杂志》“大事记”专栏的文献整理试验

Timeline Extraction of News in Historical Newspapers and Periodicals of Modern China—An Experiment on the Basis of Chronicles in Oriental Magazine

近代报刊的新闻报道记录并见证了这一特定历史时期的社会世相和众生百态,具有重要的史料价值。然而,从时间的维度,追踪和剖析海量新闻中重要历史事件的发展脉络,相关研究尚为数不多。该文由此构建针对近代报刊的时间信息标注工具,从批量新闻报道中抽取时间信息并自动转换为标准格式;在此基础上提出新闻网络模型,计算报道之间的时间关联性和内容相似性,利用有向加权网络的特有属性,追溯并探索特定历史事件的来龙去脉,不仅可以方便读者按照时间进程浏览新闻的前情回顾和后续发展;而且可以帮助研究者高效获取事件的背景知识和演变态势。该文使用《东方杂志》1911—1921年的“大事记”专栏的新闻报道作为实验数据,构建新闻网络,智能抽取历史事件的时间线,并结合特定史料加以分析佐证,旨在为近代史的知识发现提供新的研究视角。

News in historical newspapers and periodicals of modern China recorded the political and social events that took place during this time period.However,there is a notable lack of research in tracing the development of significant events in extensive historical news.Therefore,in this paper we develop a temporal tagger especially for news in newspapers and periodicals of modern China,which can help users to extract temporal expressions from news and normalize them into standard time formats.On this basis,we propose a news network model and calculate the similarities between contents and temporal relations of different news.We exploit network properties of this directed weighted network to extract the timeline of historical events.This timeline,which not only facilitates users to read news which automatically following a time-order,but also assists researchers with background knowledge and dynamic evolution of important historical events.In this paper,we use a dataset of chronicles of events from 1911 to 1921 in Oriental Magazine for experiments and corresponding experimental results can provide a new perspective for the knowledge discovery in modern Chinese history.