Abstract

In the 21st century, social media apps like Twitter, Facebook, YouTube, Waze, Instagram, etc. are used by people to post their activities, sentiments, and feedback. Likewise, people are posting their experiences either positive or negative feedback or recommendation to improve the existing services of public transportation such as Buses, Tram, Metro, Taxis, Marines. Moreover, people are also posting about traffic conditions and accidents on the social media. However, at present most of the authorities depend on surveys and customer complaints to improve to knew people sentiments. This has very limited information because they cannot reach to all customers. Therefore, in this project, we proposed a solution that utilizes Text Mining and Natural Language Processing techniques in order to determine the sentiment of the people from the collected social media data. The output of this proposed solution will be used to find the relevant information, which can add value to different sponsors within an organization such as customer service department, planning and operations to improve the existing services and optimize the current processes. We used word embedding techniques such as Word2Vec and Doc2vec to translate text into a vector format. Moreover, we developed a voting classifier in order to classify our text into three sentiment categories; negative, neutral or positive. The data for this project is crawled from Twitter using a pre-built API.

Publication Date

12-15-2019

Document Type

Master's Project

Student Type

Graduate

Department, Program, or Center

Graduate Programs & Research

Advisor

Ioannis Karamitsos

Campus

RIT Dubai

Download

Share

COinS