Skip to content

Latest commit

 

History

History
26 lines (19 loc) · 2.44 KB

project-introduction.md

File metadata and controls

26 lines (19 loc) · 2.44 KB
Raw

Project Introduction

THE PROBLEM

Too much data, no way to clearly visualise overall trend

Nowadays there is a deluge of social media data available every day. Too much data for swift accurate processing by people to determine the overall message conveyed. Brands need to understand the overall feeling towards their new products. They could read 2,000 tweets - but it would not be a good use of time. Instead a better solution would be to use a sentiment analysis AI module to analyse them and return the overall feelings towards their product. In the same way if you’re in a theme park and you’d like to quickly see which rides or parts of the park are having positive reactions. Reading through thousands of tweets relating for each ride or area would take a while. Instead a visualisation could show where positive conversations or tweets are happening as opposed to less positive reactions. There are many other times you may want to quickly gauge overall sentiment towards a topic, such as comparing opinions on two politicians running in an election, or finding the most positive time of day for a particular topic.

PROJECT AIMS

The aim of this project is to:

  • Fetch relevant results from the mass amount of social media data available
  • Automatically calculate opinions on this data set, to allow for users to gauge overall attitude towards a given topic without having to read each tweet or post
  • Plot a series of visual analytics, to find and display trends between the sentiment data and other factors such as time, location, keywords or other topics
  • To have a single centralised dashboard providing all useful data to the user in real-time and with links to further break down the results

A secondary aim of the application is to research into the area of sentiment analysis, and to develop and publish an open sauce sentiment analysis module.

BRIEF DESCRIPTION OF PROPOSED DELIVERABLE

The proposed deliverable will be in two parts:

  • A web application, publicly accessible through any modern browser. It will contain links to each of the dynamic real-time sentiment visualisations, as well as a search page where the user can enter a custom topic, and fresh tweets will be fetched, analysed and rendered.
  • A series of open source packages to do specific tasks (including the sentiment-analysis module, a fetch-tweets module, geo-lookup etc.…) Each of these will be tested, documented and then publicly published for other developers to make use of.