SEMANTIC ROLE ANNOTATION FOR ELEARNING
SEMANTIC ROLE ANNOTATION FOR ELEARNING

Summary/Abstract: The analysis of semantic roles reveals the hidden structure of a sentence and contributes to the construction of meaning, identifying specific roles that entities play in various contexts and actors involved in an event. The semantic role expresses the correlation between a predicate and its arguments. This paper describes a preliminary study about the impact semantic roles could have in eLearning contexts. The goal of our application is to identify, from a collection of learning materials, all contexts referring to a specific entity, in order to analyse relations between the entity and words with which it frequently co-occurs. Thus, through semantic role analysis, we intend to determine temporal, spatial or modal constraints which determine or restrict a concept. More concretely, the system we propose starts from an input concept, searches learning materials for that particular concept, selects the snippets that contains it, and applies semantic role labelling. At a further step, we extract semantic relations between the entity and neighbouring words, resulting in a list of binary relations. Finally, the program uses WordNet hypernyms trying to generalize over all extracted snippets. Thus, the program may facilitate the understanding of the concept through its neighbours, by creating a map of structured data related to a target concept, where each related entity is marked with its corresponding role (which can be of type Agent, Patient, Effect, Location, Cause, Time, etc.).

• Details
• Contents