Open AccessKnowledge Based System using Ontology for Accessing Sentiments of Indian Railways Tweets
Author(s) -
Rakesh kumar Donthi,
Md. Tanwir Uddin Haider
Publication year - 2019
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.j1147.0881019
Subject(s) - computer science , knowledge base , ontology , punctuality , unstructured data , schema (genetic algorithms) , data science , world wide web , information retrieval , knowledge management , data mining , big data , philosophy , statistics , mathematics , epistemology
The interactions in social networks like Twitter are reflecting sentiments of people at large. Especially opinion mining has wherewithal to provide social feedback that complements traditional feedback for making strategic decisions. The sentiments data is basically stored in the Relational Data Base. But the demerits of this data base is that it comes back with an answer at most once and it is very difficult to address complex queries over the data, lagging inherent properties such as transitivity or symmetry. It also has closed world assumption i.e., what is not known to be true in the data base is by default considered as false because knowledge represented in the data base is assumed to be complete. To overcome all these demerits, we have developed knowledge-based system using ontology, for accessing sentiments of Indian Railways tweets. Indian Railways (IR) is very huge organization which consistently strives to improve its services from time to time. In our prior work, a framework was proposed and implemented using sentiment analysis for Indian railways tweets. The results were grouped into clusters based on five attributes pertaining to IR such as Cleanliness, Staff Behaviour, Punctuality, Security and Timeliness. These clusters are being updated from time to time to reflect up to date social feedback. However, the problem with the existing system is that, its accessibility and ease of use to stakeholders is not easy. So, we proposed a knowledge based system which will represent clusters for a universal and interoperable data representation that is from database to RDF schema to ontology and apply inference rules and query using SPARQL. This system is accessible to humans and also programs in heterogeneous Machine-to-Machine (M2M) environments. We proposed a methodology to achieve this and the knowledge is made available for further processing and stakeholders can access. The proposed system is evaluated with a prototype application and found to be useful and flexibleV