During the last decade, social media information has experienced constant growth. This
fact is largely due to the increment of users using these platforms and the amount of data
and relevant information about them. Moreover, social media has completely changed the
way people communicate and interact with the rest of the world. This impact has directly
affected customer engagement. Some of the most important companies have decided to
modify their customer service model based on phone calls and paper forms to be adapted
to this new way of communicating between users and companies.
Keeping this in mind, this project aims to determine if it is possible to dissect a customer
service on the Twitter social network. Specifically, we will focus on analysing the questions
and concerns that users from the Uber transport service have. To do this, we have created a
dataset containing tweets from the English speakers addressing the @Uber Support platform
during the year 2020.
Firstly, we performed a global analysis of the collected data. This first approach helped
us to understand how users publish information in the context of both transport and customer services. Secondly, to characterize the dataset’s underlying structure, we implemented
a Topic Modelling System combined with a Topic Clustering one based on a hybridized Genetic Algorithm. The performance of both systems described a result of forming seven
groups of topics in order to cluster the different tweets from the dataset efficiently. Finally,
the combined implementation of both systems also allowed us to characterize the polarity
associated with each topic, giving us the complete stance of users towards the specific issues
described in the platform.
To conclude, we believe that the systems created in the project can be efficiently used
for the periodic analysis of the information on social media. Thus, companies who require
public opinions to establish marketing approaches can easily handle this high user-generated
content by using the systems described.