Experiments for the Extraction of Qualia from Web in Italian: a Pattern-Based Approach
Fabio Celli · 2009
This paper address the problem of automatically extracting qualia structures from the web via patterns in Italian. Qualia are treated as semantic relations, expressed by word patterns, between two entities. 400 examples of italian patterns expressing qualia are sampled and manually annotated in order to study the behaviour of the patterns and their suit-ability for qualia extraction. Features in the annotation schema include the position of the related entites with respect to the patterns, their in-formativeness and the minimum context window necessary to catch them. The results of this study are useful for designing experiments with word-space models. 1 Introduction and related work Qualia structures, originally proposed by Pustejovsky [4], are simple and pow-erful semantic structures defining the characteristics of objects in the world. The task of extracting qualia from the web has already been tried in English