RiazGirju-ICSC2010 - Another Look at Causality Discovering...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Another Look at Causality: Discovering Scenario-Specific Contingency Relationships with No Supervision Mehwish Riaz and Roxana Girju Department of Computer Science and Beckman Institute University of Illinois at Urbana Champaign { mriaz2, girju } @illinois.edu Abstract —Contingency discourse relations play an important role in natural language understanding. In this paper we pro- pose an unsupervised learning model to automatically identify contingency relationships between scenario-specific events in web news articles (on the Iraq war and on hurricane Katrina). The model generates ranked contingency relationships by identifying appropriate candidate event pairs for each scenario of a particular domain. Scenario-specific events, contributing towards the same objectives in a domain, are likely to be dependent on each other, and thus form good candidates for contingency relationships. In order to evaluate the ranked contingency relationships, we rely on the manipulation theory of causation and a comparison of precision-recall performance curves. We also perform various tests which bring insights into how people perceive causality. For example, our findings show that the larger the distance between two events, the more likely it becomes for the annotators to identify them as non-causal. Keywords-causality; contingency; scenario; topics; I. INTRODUCTION Unlike computers, people are very good at perceiving and inferring the causal, reason, purpose and explanation relationships between events in a discourse context. De- tecting such relations helps them make sense of the con- stantly changing flow of events in their daily activities and interactions. Thus, causal reasoning enables people to find meaningful order in events, which in turn helps them plan and even predict the future [13]. In linguistics, these relations form a class known as con- tingency discourse relations (cause-consequence, argument- claim, instrument-goal, purpose and reason/explanation) which are different from additive relations (list, opposition, exception, enumeration, temporal, and concession) [18], [14]. Since they are very related semantically, contingency relations can be identified as the class of causal relations in a broader sense [7]. Examples of such relations are: (1) Teachers are not going to work today because they are on strike (explanation). (2) The company makes and repairs cell phones (temporal). Thus, two events are contingent if the occurrence of one event enables the occurrence of the other – i.e., they are linked by one of the contingency relations in the dicourse context. Identifying contingency relations is very important for a number of natural language understanding tasks, such as textual entailment and explanation question answering [6]....
View Full Document

Page1 / 8

RiazGirju-ICSC2010 - Another Look at Causality Discovering...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online