Research On Interactive Timeline System Of News Report

In today's era of explosive network information,news reports have become more and more convenient to publish,and at the same time,they have been updated faster and faster.When major events occur,a large number of relevant news reports will appear on major websites.For users,how to find useful information for themselves and quickly understand the development process of the entire incident has become an urgent problem to be solved.Events that develop dynamically over time often change over time,including several different developments in the event.If there is a system that can automatically sort out the development context of the entire event,it will help users quickly grasp the development process of the incident.This article explores the process of excavating the dynamics of events from news reports that describe events,and extracts abstracts from news stories that describe the development of events.The main work of this article includes the following three points.First,a two-phase event development dynamic discovery algorithm is proposed.First,using hierarchical clustering methods,news reports are clustered according to the time of publication of news reports.In the process of hierarchical clustering,setting the threshold size is a more difficult problem.In order to solve the problem of threshold setting,a smaller threshold is set in the process of hierarchical clustering so as to obtain the susceptibility development trend of the event.However,there are problems with the division of fines in the development of suspected developments.In order to solve the problem that the development of the suspected development is too finely divided,the modularity function is used as the objective function,and clustering is performed based on the text similarity for the suspected development dynamics,so that the similarity between documents in the same class is greater than the similarity between classes and classes.The class obtained after clustering the suspected development dynamics is the dynamic development of the event,thus solving the problem of threshold selection.Second,the Lex-PageRank algorithm is proposed to extract the overall abstract of all documents in the dynamic development of an event.The algorithm not only considers the mutual influence of words in sentences and sentences in the document,but also considers the influence of the title on the sentences and words.The Lex-PageRank algorithm allows words and sentences to be voted against each other to obtain the importance of sentences and words in the document and general weights.Finally,extract the important sentence from the document according to the redundancy between the sentence score and the sentence.Thirdly,the two algorithms described in the first and second applications are used to improve and implement the timeline generation module in the TimeLineCurator system.The function of the improved timeline generation module is that when the user submits a news report describing the event,the module is responsible for events.The development of each of the developments is carried out,abstracts are extracted for each development,and finally the development of the events is shown in a visual timeline.