Visualizing emotions in social conversations
Social media comments are a valuable resource for identifying sentiments and opinions that are posted by commenters across geographical and socio-economic backgrounds. In this report we illustrate how we apply sentiment analysis on YouTube comments with the help of our existing KANDINSKY Mobile appl...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2022
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/157546 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Social media comments are a valuable resource for identifying sentiments and opinions that are posted by commenters across geographical and socio-economic backgrounds. In this report we illustrate how we apply sentiment analysis on YouTube comments with the help of our existing KANDINSKY Mobile application, which provides a novel visualisation of YouTube comments, grouping related comments together as concentric circles in a virtual space. We first examine the structure of the application, followed by a detailed exploration into various popular sentiment analysis tools, particularly Sentiwordnet, SenticNet and VADER, which we then use to identify the polarity of sentiments contained in comments on YouTube videos. This is dependent on not only the words used in the sentence but also on various non-alphabetical factors such as emojis, foreign characters and exaggerated spellings that are common in social media comments. Finally we visually depict the polarities of these sentiments in the application. We hope that this project will be of great interest to academic researchers and the general public alike, as the identification and visualisation of the various types of sentiments is of significant relevance to everyday social conversations. |
|---|