farosato/sentweements

Final project for Data Mining course of M.Sc. in Engineering in Computer Science at Università degli Studi di Roma "La Sapienza" (A.Y. 2016/2017), developed in collaboration with Giacomo Lanciano.

Final project for Data Mining course of MSc in Engineering in Computer Science at Università degli Studi di Roma "La Sapienza" (A.Y. 2016/2017).

The project

Sentweements is a beautifully intuitive sentiment analysis tool for tweets.

Using Twitter Streaming API to fetch tweets in real-time and Indico's artificial intelligence APIs to perform textual sentiment analysis, data is visualized on a beautiful, interactive choropleth map showing Italy's sentimental situation.

With Sentweements you have:

• national and region-specific statistics, available through mouse hover.
• data persistence implemented with SQLite DB.
• dynamic updates automatically reflected on the map as new data becomes available (Twitter streaming data).
• static analysis mode available by defining a time window of tweets to analyze, to get different perspectives.
• api keys carousel and multi-threaded data retrieval architecture to expand rate limit both for Twitter and sentiments APIs (check secret_keys_template for instructions).

See also our images streaming emotion analysis.

Technologies

• Indico - an artificial intelligence service that detects sentiments in texts, available for several different languages.
• Twitter Streaming API - real-time access to tweets coming from all over Italy.
• Flask - Python microframework to build the webserver and serve client requests.
• Leaflet - Javascript framework to build the choropleth map visualization.

To run

Be sure to have Python 3 installed.

Note that some additional Python modules are required; you can run $ python dependencies.py to install them all in one shot.

To start tweets retrieval, open a terminal and type $ python tweets_streaming.py.
Note that you need a working Internet connection to download the tweets from Twitter.

To run the webapp (locally), open a terminal and type $ python webapp.py.
Then, open a browser and go to localhost:5000.

Developers

Fabio Rosato	[email protected]
Giacomo Lanciano	[email protected]