David-O-Meter

Project page moved. Click here

What is this?

         This is a program/webpage that displays the ratio of positive to neutral to negative tweets that mention the word "David".
         The David-O-Meter measures the overall sentiment toward Davids around the world  :)

How does it work?

         The java program first sends a request to the Twitter API to return recent tweets that mention the word "David" from Twitter users not named David.
         Then the returned JSON data is parsed and the tweets are formatted into a format recognizable by Azure Cognitive Services. The new JSON containing the tweets' text content is sent to the Azure Cognitive Services API for sentiment analysis.
         The returned data is then parsed again, and any sentiment with a .5+ confidence score is added to its' aggregate count. For example, a tweet with a positive score of .6, negative of .15 and negative of .25 would increase the positive sentiment count by 1
         Twitter limits me to 300 API requests per 15 minutes so this process is run every 3 seconds to have it repeat no more than 300 times in 15 minutes. Every 15 minutes, the sentimentValues.json file—which contains the total values of each sentiment—is updated, commited, and pushed to the repo
         When a client visits the webpage, the initial HTML/CSS will have each sentiment in 1 of 3 bars in a flex container, all with equal lengths and content. The JavaScript will then parse the values from sentimentValues.json and update each sentiment's flex-grow property (to create the ratio in length) and its content/innerHTML to display the percentage for that sentiment.

Why did I do this?

         I wanted to learn about APIs, how to use them and how to use the data extracted from them in client-facing software/sites. So, instead of buying an expensive course to slowly learn it, I decided to think of a simple non-controversial project to work on and learn across the way with the help of documentation (Twitter API documentation sucks though) and Google (I actually use Bing).

Am I finished?

Short answer: No
Long answer: While the project and site are functional, here's some features and stuff I wish to add soon:

• Ability to see sentiment values for the last hour and last 24 hours
• Ability to see the recently analysed positive, neutral and negative tweets
• Ability to download program software as jar and deploy it to analyze and display sentiment for whatever key word/phrase you choose