Silvia Pareti

Annotating attribution relations and their features

Attribution has attracted increasing attention in recent years following the development of Opinion Analysis and Information Extraction applications. Given a piece of information or an opinion, being able to correctly identify its source (either a specific entity e.g. President Obama or a class thereof e.g. experts, official sources, rumours) and determine how different sources and features of attribution affect the way information is perceived would be extremely beneficial. It would in fact enhance opinion-oriented applications of Language Technology and revolutionise the way we can select information, e.g. on the basis of source expertise and reliability. However, approaches to the automatic extraction of attribution relations have achieved only partial results to date and are therefore not sufficient to enable the development of reliable applications. In addition, there has been little or no attempt to discern the impact of different sources and attribution strategies on the perception and interpretation of the attributed material. Grounded on previous analysis of this relation in Italian [2] and English [4] news corpora, this study addresses several of the attribution strategies identified to develop a more comprehensive system for the extraction of attributions and their relevant features from news texts.