Semantic Relation Extraction from Legislative Text Using Generalized Syntactic Dependencies and Support Vector Machines

In this paper we present a technique to automatically extract semantic knowledge from legislative text. Instead of using pattern matching methods relying on lexico-syntactic patterns, we propose a technique which uses syntactic dependencies between terms extracted with a syntactic parser. The idea is that syntactic information are more robust than pattern matching approaches when facing length and complexity of the sentences. Relying on a manually annotated legislative corpus, we transform all the surrounding syntax of the semantic information into abstract textual representations, which are then used to create a classification model by means of a standard Support Vector Machine system. In this work, we initially focus on three different semantic tags, achieving very high accuracy levels on two of them, demonstrating both the limits and the validity of the approach.