Argument extraction for supporting public policy formulation

摘要

In this paper we describe an application of language technology to policy formulation, where it can support policy makers assess the acceptance of a yet-unpublished policy before the policy enters public consultation. One of the key concepts is that instead of relying on thematic similarity, we extract arguments expressed in support or opposition of positions that are general statements that are, themselves, consistent with the policy or not. The focus of this paper in this overall pipeline, is identifying arguments in text: we present and empirically evaluate the hypothesis that verbal tense and mood are good indicators of arguments that have not been explored in the relevant literature.