Monitoring the Quality and Perception of Service in Colombian Public Service Companies with Twitter and Descriptive Temporal Analysis

The main goal of this research is to analyze the perception of service in public sector companies in the city of Bogota via Twitter and text mining to identify areas, problems, and topics aiming for quality service improvement. To achieve this objective, a structured method for data modeling is implemented based on the KDD methodology. Tweets from January to June 2022 related to the companies in the sector are processed, and a temporal analysis of the evolution of sentiment is performed based on the dictionaries Bing, AFINN, and NRC. Subsequently, the LDA algorithm (Latent Dirichlet Allocation algorithm) is used to visually identify the topics with the greatest negative impact reported by the users in each of the 6 months by adding the temporal dimension. The results revealed that, for Aqueduct (water supply service), the topic with the highest dissatisfaction is related to the Water Tank Request processes; for Enel (energy services) Service Outages; and for Vanti (gas services), Case solution and request information. Temporal patterns of tweets, sentiments, and topics are also highlighted for the three companies.

Automated summary

The research aims to analyze service perception in public sector companies in Bogota using Twitter and text mining. By implementing data modeling and analyzing sentiment evolution, the study identifies areas, problems, and topics for improvement. The LDA algorithm helps visualize the most negatively impactful topics reported by users over different time periods.