Identification and Prioritization of urban issues from Smart City data

Government should be able to publish and interact better for offering e-government services effectively to their netizens. In this context publish refers to delivering information through web pages and interact refers to allowing citizens to provide feedback and suggestions. Developing a model for understanding and prioritizing urban governance issues from these user-generated contents is an important research issue. This paper proposes a four-step solution approach to prioritizing urban governance issues from user-generated contents using Aspect Based Sentiment Analysis (ABSA). First, a novel lexicon-based approach is used for selecting trigrams as aspect phrase and pruning the list using cross-domain stop words. Second, the opinionated sentences are extracted using the identified aspects. Third, polarity and subjectivity of the aspects and corresponding comments along with the bag-of-words for the aspects are considered as the features for learning the category of the urban issue. Fourth, the sentiment score of a specific category is used to prioritize the urban issues. The proposed methodology is applied to a real user-generated content extracted from a discussion forum on smart cities from an e-governance portal in India. Our research demonstrates that Aspect Based Sentiment Analysis (ABSA) with necessary modifications can be utilized to extract issues, suggestions, and ideas from crowd-sourced citizen-generated content (Social Data). Sentiment analysis can be used to prioritize them as well. Analysis of the data for the candidate city shows that while the citizens are happy about educational facilities in the city they are highly concerned about health issues. A deeper investigation revealed the health issues are due to high degree of carbon emission, lack of solid waste management system and lack of sanitation among the others.