Twitter Sentiment Analyzer Service

A containerized service for inferring the sentiments of tweets by topic.

Code and architecture

The implementation is explained on code_description.md.

Service usage

The service was deployed on heroku csi02-tsas.herokuapp.com.

There are six routes:

• /api/list: returns a list of tweets for the #1 Trend Topic with their sentiment analysis.
• /api/list/<topic>: returns a list of tweets for the given topic with their sentiment analysis.
• /api/sumlist: returns a list of tweets for the #1 Trend Topic with their sentiment analysis and a summary of that sentiment analysis.
• /api/sumlist/<topic>: returns a list of tweets for the given topic with their sentiment analysis and a summary of that sentiment analysis.
• /api/summary: returns just a summary of the sentiment analysis of the #1 Trend Topic.
• /api/summary/<topic>: returns just a summary of the sentiment analysis of the given topic.

The complete API documentation is available at https://csi02-tsas.herokuapp.com/.

Running the service

⚠️ You will need a Twitter API key placed in a secrets/twitter_api_key.json file like below:

{
  "API_KEY": "<your twitter api key>",
  "API_KEY_SECRET": "<your twitter api key secret>"
}

⚠️ You will also need to generate your own Google Cloud key and put the resulting json file in the secrets folder as well.

1. Build

docker build -t tsas/csi-02-tsas .

2. Run

Without cache:

docker run -p 8000:8000 -ti tsas/csi-02-tsas

Saving cache:

docker run -p 8000:8000 -ti -v `pwd`/caches:/usr/src/app/caches tsas/csi-02-tsas

So you can access the service locally on localhost:8000.