Discovering Event Evolution Graphs Based on News Articles Relationships

Abstract

There are many news articles reported online everyday. Within an ongoing topic, people can find a huge amount of news articles. A topic often consists of several events, and people are interested in the whole evolution of a topic along a timeline. This requests for finding and identifying the dependent relationships between events. In order to understand the whole evolution of a topic effectively, we propose a framework of event relationship analysis. We define three kinds of event relationships which are co-occurrence dependence relationship, event reference relationship, and temporal proximity relationship for modeling how an event is dependent on another event within a topic. Through combining three kinds of relationships, we can discover an Event Evolution Graph (EEG) for users to view the evolution of a topic. Experiments conducted on a real data set show that our method outperforms baseline methods.

Event-based summarization of news articles

“…News sites such as CNN, 1 Rappler, 2 and BBC 3 inform users about current events. An event can contain several events or other dependent events, which indicates a semantic relationship among them [45]. Therefore, in this research the relationships between events and event semantics are investigated to extract the meaningful summaries as presented in the EBDS framework in the Figure.…”

An uncertain future: Predicting events using conditional event evolutionary graph

“…Lu et al 22 applied multiple similarities to make a fine‐grained analysis of the entire event relationship. Furthermore, Huang et al 23 used three kinds of event relationships, including co‐occurrence dependence analysis, event reference analysis, and temporal proximity analysis for modeling how event evolve from one to another.…”

“…The transition probability will be replaced if P ( b i | a , c′ ) > P ( b i | a , c ) when the context of event changes. EEM : Yang et al 5 apply content similarity, temporal proximity, and document distributional proximity to measure the relation between events. As it is difficult to extract time accurately, we only use content similarity and distributional document proximity as the relationship measurement and take the annotated direction as the evolution direction of the method. NAR : Huang et al 23 use co‐occurrence dependence analysis, event reference analysis, and temporal proximity analysis to measure the event relationship. We use co‐occurrence dependence analysis and event reference analysis to analyze the relationship between events, and take the annotated direction as the evolution direction of the method. Method without general context (EEG) : We only calculate the probability of an event and ignore the generalization context.…”

Challenges of event detection from social media streams

“…Still, this does not satisfy the news readers nowadays as they are not just interested in detecting the significant events, but also in how these events have evolved along the timeline [56]. However, fulfilling this requirement has proven to be a very challenging task, especially for high-level rate stream [18]. On the other hand, the Feature Pivot approach depends on detecting burst features from SN text streams and focuses on the variation of detected features [1].…”

“…That is due to the dynamic characteristics of the social media as well as the existence of both text content and network structure within the streams [11] . Thus, the process of information filtering, analyzing data, and especially, detecting and monitoring the interesting events from social media text, has become the most difficult task [18]. In the following paragraphs, the existing challenges caused by SNs for ED are divided into two categories: General text mining challenges for ED from SNs; and specific challenges of the ED methodology.…”