Concept Identification from Poems

Abstract

Emotion recognition has become an important topicin natural language processing. Usually words labeled with theiremotion is the starting place to find the emotion of a text, but itis very essential that context must also be considered. It is not asimple task to capture the overall emotion of a Text, as words aremutually influence their emotion related interpretation. There isa lot of word-based dictionaries available for emotion recognitionfrom the text. However, commonsense knowledge bases arevery less. We used ConceptNet (commonsense knowledge) forconcept mining from poems. In this paper we proposed a conceptidentification method by using ConceptNet knowledge base todepict the concept of poems. This system gives a precision valueof 71% and inter-rater agreement of 48%, depicting its moderateagreement with Human Expert interpretation.