ITSA is a tool for collecting tweets from specific twitter users and applying sentiment analysis to the tweets. This analysis is applied in three stages through a CLI. The results are stored in CSV files for further analysis.
It is recommended that you place the dependencies in a folder called lib, otherwise you will need to change some code for SentiStrength.
- Twitter4J 4.0.4
- Apache Commons CSV 1.4
- SentiStrength
- SentiStrength Dec 2015 Data - place in
lib/sentistrength_dataor modify the code accordingly - SentiStrength jar
- SentiStrength Dec 2015 Data - place in
This program goes through 3 stages. Through the CLI you may select which stages to run. The results from each stage are stored in CSV files
These stages are as follows:
- Data collection
- Tweets are collected from twitter from the users defined in
itsa.properties
- Tweets are collected from twitter from the users defined in
- Normalization
- Tweets are stripped of their links
- Sentiment analysis
- Positive and negative scores are calculated for each tweet
Those interested in continuing the work of this project should read the Future Work section of the paper in the data and paper zip file. Here are a few relating to implementation:
- The twitter API limits the number of tweets that can be collected from a user to 3200
- No modularity in first phase, must be done manually
- Limited configuration options. Most tweaks will need to be made in the code.
- Access to CSV fields is done by index (0, 1, 2... etc). Doing it by header would mitigate many issues