That's very broad question, Poria.

What do you want to do with your thesis? I can quickly think of several different directions:

• demonstrate the skills to run a sentiment analysis using existing methods and technologies
• review existing methods and recommend the "best" one
• take an existing method based on Python and reproduce it in Julia so that it's faster
• create an R package that lets a user quickly access Instagram feeds and topics
• construct, or improve on, a vocabulary for Twitter analysis in a language other than English
• test alternative classification algorithms to find options that are quicker or more accurate than the usual Naive Bayes method.

Each of these topics requires different skills and different tasks. Some may require you to gather new data through social media. Others can be done with existing data curated on Kaggle or other sites. Some require higher-level programming.

What are the expectations of your Masters thesis? If it's a project within a mostly course-work Masters degree, then a demonstration of competence using an existing technology is probably sufficient. If this project is the Masters degree then you may need to do something more substantial.

Talk to potential supervisors.

You might find these works helpful:

Conference Paper Sentiment Prediction in Social Networks

Conference Paper Emotions in Social Networks: Distributions, Patterns, and Models

Hi.,

The Complex Event Processing (CEP) is a methodology that combines data from many ... The SA is a tool which assists the shifter in his/her daily work, diagnosing problematic ... It works by providing checks and evaluating conditions in real time, ... sandbox, using a configurable replica of past and archived data, events ..

For More info:

Article An open architecture for event-based analytics

https://dl.acm.org/citation.cfm?id=1142520