NEW: Accepted papers
The 4th Workshop on EVENTS will be held at the NAACH HLT 2016 conference in San Diego, California June 16 or 17, 2016.
Proper understanding of the events of a text is increasingly important in NLP. However, there is no clear consensus regarding the definitions, representation, coreference or relations between events. We seek to bring together researchers focused upon these wide-ranging topics, to bring different treatments of events into harmony.
The definition and detection of events largely come down to questions of which spans participate in events and to questions about which events should be annotated at all. Many annotations of event coreference have been limited to specific scenarios or domains, as in LDC’s ACE and Machine Reading event annotation (Humphreys et al. 1997; Bagga and Baldwin 1999; He 2007). Domain-specific detection of events has approached human performance in the recent Clinical Tempeval evaluations (Bethard et al. 2015). However, despite their similarities, detection of events within ontologies has largely remained a separate endeavor from general event detection.
Similarly, dramatic strides have been taken in event coreference. The OntoNotes corpus annotations include general event mentions and coreference, but mainly identify coreference between verbs and nominalizations (Pradhan 2007).
Bejan and Harabagiu (2010) have offered broader event coreference annotation for evaluation purposes, which have been revised and extended by Lee et al. (2012) and Cybulska and Vossen (2014). More recently, event coreference resolution systems were presented by Liu et al (2014) and Araki and Mitamura (2015). Less consensus, however, exists in bridging anaphora and subevent relationships between events (Poesio and Artstein 2005; 2008, Recasens et al. 2011, Araki et al. 2014).
Relationships between events, and that events have with scripts, is another major issue related to events. TimeML (Pustejovsky, et. al., 2010), or temporal relation annotation, has held numerous Semeval tasks regarding detection of temporal relations. Other relations between events, such as causation (Bethard and Martin 2008; Mizra et al. 2014) or contradiction detection (De Marneffe et al. 2008), have been studied, some finding great value in event coreference. However, outside of temporal relations there is a paucity of consensus regarding which phenomena to study and a lack of shared evaluations regarding them.