Argument Mining: the Bottleneck of Knowledge and Language Resources

Argument Mining: the Bottleneck of Knowledge and Language Resources Patrick Saint-Dizier IRIT-CNRS, 118 route de Narbonne, 31062 Toulouse cedex France, stdizier@irit.fr Abstract Given a controversial issue, argument mining from natural language texts (news papers, and any form of text on the Internet) is extremely challenging: domain knowledge is often required together with appropriate forms of inferences to identify arguments. This contribution explores the types of knowledge that are required and how they can be paired with reasoning schemes, language processing and language resources to accurately mine arguments. We show via corpus analysis that the Generative Lexicon, enhanced in different manners and viewed as both a lexicon and a domain knowledge representation, is a relevant approach. In this paper, corpus annotation for argument mining is first developed, then we show how the generative lexicon approach must be adapted and how it can be paired with language processing patterns to extract and specify the nature of arguments. Our approach to argument mining is thus knowledge driven Keywords: Argument mining, Knowledge representation, Generative lexicon 1. Aims and Challenges of Argument Mining One of the main goals of argumentation mining is, given a controversial issue, to identify in various texts the argu- ments for or against that issue. These related arguments act as supports or attacks of the issue, depending on their ori- entation. Arguments may also attack or support the other arguments which support or attack that controversial issue in order to cancel out or reinforce their impact. An argu- mentation is represented by a graph of supports and attacks. Arguments are difficult to identify and to characterize, in particular when they are not adjacent to the controversial issue, possibly not in the same text. Arguments are often standard natural language statements which get the status of arguments because of the specific relations they have with a controversial issue. For example, the statement from the Indian Times: (a) we now see long lines of young girls with school bags along the roads could be just factual, but given the controversial issue: (b) the situation of women has improved in India, then, statement (a) becomes an argument supporting (b), which is then interpreted as a controversial issue which can then be supported or attacked. Except in specific contexts, and for certain forms of arguments (e.g. warnings, threats, advice, requirements), most arguments do not have any spe- cific linguistic mark that would allow to identify their rela- tion to a controversial issue. Furthermore, it is difficult to identify whether a statement is a support or an attack of a controversial issue, and what it precisely attacks or sup- ports. In the above example, school bags means education: it is a means to improve women’s condition because it leads to jobs and more independence. It does not talk about any- thing else concerning women’s conditions. Relating a con- troversial issue with arguments requires knowledge, lexi- cal semantics data, and appropriate inferential patterns. It is therefore much more complex in terms of semantics and reasoning than e.g. …

Content truncated to 500 words