Crowdsourcing Salient Information from News and Tweets

Crowdsourcing Salient Information from Tweets and News Oana Inel*, Tommaso Caselli, Lora Aroyo The Network Institute,Vrije Univerisiteit Amsterdam; *IBM Nederland, Center for Advanced Studies Benelux De Boelelaan 1105 1081 HV Amsterdam; Johan Huizingalaan 765, 1066 VH Amsterdam {oana.inel,t.caselli,lora.aroyo}@vu.nl Abstract The increasing streams of information pose challenges to both humans and machines. On the one hand, humans need to identify relevant information and consume only the information that lies at their interests. On the other hand, machines need to understand the information that is published in online data streams and generate concise and meaningful overviews. We consider events as prime factors to query for information and generate meaningful context. The focus of this paper is to acquire empirical insights for identifying salience features in tweets and news about a target event, i.e., the event of “whaling”. We first derive a methodology to identify such features by building up a knowledge space of the event enriched with relevant phrases, sentiments and ranked by their novelty. We applied this methodology on tweets and we have performed preliminary work towards adapting it to news articles. Our results show that crowdsourcing text relevance, sentiments and novelty (1) can be a main step in identifying salient information, and (2) provides a deeper and more precise understanding of the data at hand compared to state-of-the-art approaches. Keywords: crowdsourcing, event salience, event topic space 1. Introduction The key feature of the current information age is the contin- uously growing stream of information of various types, e.g., news, tweets, videos, and across various distribution chan- nels, e.g., social, traditional media, personal blogs, infor- mation portals. All this results in an inconceivable amount of information redundancy, i.e., the same item is re-shared across different channels in identical or similar forms. This poses challenges both for humans and machines. Humans need to reduce the information overload and to be able to identify the truly novel and relevant information items. Machines need to generate meaningful news clusters (by means of news aggregation systems1,2) by identifying the novel information items that are relevant to target stories. Notions like novelty, relevance, and salience play a central role when dealing with such huge and continuous informa- tion streams. Defining these notions is not a trivial task as they are deeply interconnected. We consider salience, or in- formation salience, in the context of relevance and novelty and we adopt the following definitions: • novelty: corresponds to any bit of (relationally) new information. From this point of view, we follow the TREC definition of novelty, that is something which is presented as new with respect to a given context, which corresponds to the known information. • relevance: identifies “something important”. Impor- tance can be determined only in relation with some- thing (e.g., an object, a topic, an event, among others) and in a context of occurrence. Subjectivity and in- tentionality play an important role in determining the relevance of an information item. • salience: combination of novel and relevant informa- tion, seen as an incremental and gradable …

Content truncated to 500 words