Sentiment Analysis Using Common-Sense and Context Information

Algorithm 1

Build domain specific ontology from common-sense knowledge base.

INPUT Raw Assertions related to domain extracted from ConceptNet.

OUTPUT Ontology with domain-concepts

Step  1. Every relation r in the ontology is constructed by connecting two concepts i.e. concept1 (c1) and concept2 (c2).

Step  2. Generate a graph structure using these relations. Root of this graph is the domain itself.

Step  3. We connect two vertices V1 (i.e. concept1) and V2 (i.e. concept2) with an edge E (i.e. relation r). Connect all the

nodes extracted from ConceptNet to construct the ontology.

Step  4. First level nodes of this ontology are considered as new domain names and further synonyms are extracted from the

WordNet for expansion of the ontology.

Step  5. Repeat Steps  1–3 to construct ontology for each synonym word of the main domain.

Step  6. Merge all the extracted ontology to generate a single domain specific ontology.