This document proposes a system to predict brand popularity using opinion mining of social media and news sources along with statistical analysis. The system collects text data from subscribed keywords on Twitter and news sources, sends it to a sentiment analysis server, separates the text by location, assigns rankings to locations based on sentiment analysis and statistics, and maps and alerts users to changes in sentiment over time. It is designed with a microservices architecture and cloud-friendly design for scalability. Potential applications include political campaigns, company reputation monitoring, and product launches.
2. Introduction
•Everybody needs a brand
•Important to know brand value
•Statistical tools coupled with opinion mining are powerful
•Leverage power of social media
4. Flow
1. User subscribes to the keywords for which he/she needs popularity (Ex.
Elections, Hillary)
2. Twitter and News streams are opened for those keywords.
3. Text collected is sent to a remote opinion analysis server.
4. Positive, Negative or Neutral opinion is collected from the text.
5. Flow Cont..
1. Text is separated according to location it has come from.
2. Ranking is assigned to the location based on the text of that location.
This is done after performing Statistical functions on data.
3. User can view map of the locations and corresponding opinions.
4. Mail is sent to user if the opinion of a location changes drastically.
6. Cloud friendly architecture
Microservices architectural pattern
No tight coupling in subscribers, Core engine and Publishers
Can avail cloud PaaS, container technologies
Out of box scale out architecture and design
Core
Engine
Subscribers
- Twitter
- Facebook
- News Articles
Publishers
- Email
- App Notification
- MMS
7. Software Stack
Web UI
( HTML5,
JS)
Midldlware
Server
( Nodejs)
Subscribers
(Nodejs)
Publishers
(Python)
8. Applications
Campaigns
◦ Political, NGOs
Reputation in market
◦ Companies, Organizations
◦ Product launches, portfolios
◦ Public Figures, Celebrities
Helpful to PR guys
Stock market too ( Back to the Future with almanac )?
Unlimited possibilities
10. <source.js>
Its hackathon – You can’t prototype without open source tools, libs and
packages
Thanks to all awesome devs for libs, packages and cryptic docs :)
Source Code repository : https://github.com/aniruddh-a/rta