Extracting Semantic Entities and Events from SportsTweets

Making Sense of Microposts 2011.

计算机科学;

Smitashree Choudhury1 ; John G. Breslin2

Others  :  http://ceur-ws.org/Vol-718/paper_17.pdf
PID  :  42784

学科分类：计算机科学（综合）

来源: CEUR

【 摘 要 】

Large volumes of user-generated content on practically every major issue and event are being created on the microbloggingsite Twitter. This content can be combined and processed to detectevents, entities and popular moods to feed various knowledge- intensive practical applications. On the downside, these content itemsare very noisy and highly informal, making it difficult to extract senseout of the stream. In this paper, we exploit various approaches todetect the named entities and significant micro-events from users’tweets during a live sports event. Here we describe how combininglinguistic features with background knowledge and the use of Twitter- specific features can achieve high, precise detection results (f-measure= 87%) in different datasets. A study was conducted on tweets fromcricket matches in the ICC World Cup in order to augment the event-