Context Dependent Claim Detection

Proceedings of COLING 2014, the 25th International Conference on Computational Linguistics: Technical Papers, pages 1489–1500, Dublin, Ireland, August 23-29 2014. Context Dependent Claim Detection Ran Levy Yonatan Bilu Daniel Hershcovich Ehud Aharoni Noam Slonim IBM Haifa Research Lab / Mount Carmel, Haifa, 31905, Israel {ranl,yonatanb,danielh,aehud,noams}@il.ibm.com Abstract While discussing a concrete controversial topic, most humans will find it challenging to swiftly raise a diverse set of convincing and relevant claims that should set the basis of their arguments. Here, we formally define the challenging task of automatic claim detection in a given context and discuss its associated unique difficulties. Further, we outline a preliminary solution to this task, and assess its performance over annotated real world data, collected specifically for that purpose over hundreds of Wikipedia articles. We report promising results of a supervised learning approach, which is based on a cascade of classifiers designed to properly handle the skewed data which is inherent to the defined task. These results demonstrate the viability of the introduced task. 1 Introduction The ability to argue in a persuasive manner is an important aspect of human interaction that naturally arises in various domains such as politics, marketing, law, and health-care. Furthermore, good decision making relies on the quality of the arguments being presented and the process by which they are resolved. Thus, it is not surprising that argumentation has long been a topic of interest in academic research, and different models have been proposed to capture the notion of an argument (Toulmin, 1958; Freeley and Steinberg, 2008). A fundamental component which is common to all these models is the concept of a claim (or conclusion). Specifically, at the heart of every argument lies a single claim, which is the assertion the argument aims to prove. Given a concrete topic, or context, most humans will find it challenging to swiftly raise a diverse set of convincing and relevant claims that should set the basis of their arguments. The purpose of this work is to formally define the challenging task of automatic claim detection in a given context, to outline a preliminary solution to this task, and to assess its performance over annotated real world data, collected specifically for this purpose. In his classical argument model, Toulmin defined a claim as a conclusion whose merit must be estab- lished (Toulmin, 1958). Since we are interested not in detecting claims in general (Mochales Palau and Moens, 2009; Teufel, 1999), but rather in detecting claims that are specifically relevant to a pre-defined concrete context, we suggest a definition with a more functional flavor. In practice, we found this def- inition easy to convey to human labelers, and consequently feasible to capture by automatic detection methods. In particular, we define the following two concepts: • Topic – a short phrase that frames the discussion. • Context Dependent Claim (CDC) – a general, concise statement that directly supports or contests the given Topic. Given these definitions, as well as a few more detailed criteria to reduce the variability in the …

Content truncated to 500 words