| In the era of intelligent media,the deep development of text mining has accelerated the upgrading change of online media content provision and dissemination methods.In recent years,with the diversified exploration of news information analysis,storyline generation has become a research hotspot in the field of text mining.Storyline generation focuses on how to automatically extract and organize events from massive news data to present their inner connections,which is especially important in applications such as information retrieval and opinion analysis.However,the different narratives of the various stages of hot news events can lead to a lack of accuracy in capturing current topics in the storyline,and in addition,the neglect of a large number of derivative and transitive events in the evolution can lead to a poor integrity of the storyline.To address the above issues,this paper proposed a hierarchical storyline generation method(HSGM)for hot news events,and designed and implemented a hierarchical storyline analysis system.(1)Aiming at the problems of lack of accuracy in capturing current topics by existing storyline and poor integrity of the storyline,a hierarchical storyline generation method(HSGM)was proposed.It included a hot word extraction module based on multi-feature,an event extraction module,and a hierarchical storyline construction module.Among them,the hot word extraction module based on multi-feature extracted hot words at each stage by continuously tracking the development process of events,which enhanced the accuracy of the storyline in capturing the current topic.The event extraction module served as an intermediate stage,making the information obtained by the former available to the hierarchical construction module.The hierarchical storyline construction module utilized the guiding effect of hot words in storyline and combined a storyline coherence selection strategy and a hatchery strategy to enhance the strength of parent-child event connection,which laid the foundation for the construction of a complete storyline.Based on the above three modules,a complete hierarchical storyline generation framework was designed,which enriched the expression of the storyline by incorporating hot words into the hierarchical storyline construction,thus enhancing the accuracy and completeness of the storyline.The experiments on three real datasets in this paper showed that HSGM had improved the F-index of event tracking and the accuracy,comprehensibility,and completeness evaluation based on user experience of storyline construction.(2)In this paper,we designed and implemented a hierarchical storyline analysis system.The system was mainly divided into three modules: news search,data management,and storyline analysis.Among them,the news search module facilitated users to inquire hot events in a one-stop manner,the data management module simplified news data collection and collation,and the storyline analysis module would help users to understand hot events faster and better through visualization technology.The system simplified the tedious work of news collection and organization,and assisted news users in public opinion analysis and posture judgment. |
PDF Full Text Request