Harvesting railway passenger opinions on multi themes by using social graph clustering

• Published in: Journal of rail transport planning & management 2020-05, Vol.13, p.100151, Article 100151
• Main Authors: Myneni, Madhu Bala, Dandamudi, Rohit
• Format: Article
• Language: English
• Subjects: Passenger opinions; Railway passengers; Railway services; Sentiment analysis; Text mining
• ISSN: 2210-9706; 2210-9714
• DOI: 10.1016/j.jrtpm.2019.100151

The largest service provider in public transportation like railways has a lot of challenges with changing dynamics. Identification of heterogeneity in perceptions of service quality among groups of railway passengers to make indicators has a lot of importance in service oriented organizations like railways. Railway passengers' opinions have also been given significant importance for monitoring, enhancing the existing services, and understanding the current needs of the people. The existing practice involves people for collecting the data, along with responding to the complaints and forwarding these to the concerned heads. Now the research is centered on dynamic passenger behaviour analysis by using social data, such as twitter, the facebook etc. In this paper, the importance of dynamic social sentiment analysis and affective computing in railways, issues associated with data collection and processing are elaborated. The proposed work offers a framework for social sentiment analysis on railway passenger tweets. Firstly, it identifies the theme wise data and then provides a sentiment score which expresses about multiple themes. A novel social graph clustering approach is utilized to separate the data according to theme and perform the sentiment analysis on each cluster to anticipate the passenger opinions. This work will give the indicators and assessment of railway services quality with an understanding of the passengers' opinions. It acts as an expert feedback, service indicator, and is also helpful to enhance the services through a thorough analysis of Indian railways passengers' opinions. •Importance and present need of Affective computing and sentiment analysis on micro blog data.•In service sector, the need of an automated service indicator is elaborated.•To process the multi theme data, the solution is provided through social graph clustering.•Automate the sentiment analysis, Machine learning is used.