Semantic network for knowledge representation

A semantic network is the representation of knowledge in the form of a graph where the nodes represent entities or events and the edges connecting them describes their semantic relationship. This project aims to develop a program which can present this relationship in terms of the strength of the link, as well as analyze the emotions expressed between them. Representing knowledge in this manner presents the possibility to predict possible behaviors or attitudes exhibited by one entity towards another in a social network. The implementation involves examining a text to extract named entities and the sentiments experienced between them, as well as measure the strength of the relationship between these entities by adapting link discovery algorithms used for web-based resources. Results were obtained from evaluating sentences containing an emotive verb, and the system was shown to be ineffective in interpreting the sentences. Comparisons were also made between three connection measurement methods, and the lexical co-occurrence method was proven to be better at establishing relevant connections compared to path-based measures. The two path-based measures were tested on their ability to rank entity pair relationships, but the results were inconclusive due to implementation difficulties. Several recommendations for improving the information extraction capabilities of the program were suggested at the end of the report.Bachelor of Engineerin