Sentiment Score Plugin

Use the Sentiment score plugin to find keywords in social media input messages and flag them.

Social media has created many new opportunities, but the sheer volume of messages can be difficult to keep up with. The Sentiment score plugin provides the ability to scan messages and check for keywords, and then flag messages that are found further investigation.

 The Sentiment score plugin uses a list of emotive words, contained in four lists:

1. AFINN-111: this list contains 2477 words and assigns a score from -5 to +5 for each of these words.
2. Anew: this list contains 276 words and assigns a score from -2 to +1 for each of these words. 1=Positive; 0= Neutral; -1=Negative; -2=Anxious.
3. Lexicon: this list contains 6788 words and assigns the score 1 and -1 for each of these words. 1=Positive; -1= Negative.
4. Sentiments: this list contains 8196 words and assigns a score from -1 to 1 for each of these words. 1=Positive; 0=Neutral; -1=Negative.

The plugin matches words in the input message and assigns a new integer field with the sum of the word score in the message.

The score will probably represent the overall sentiment of the message, with a positive score for a positive sentiment and vice versa. However, there are many cases where the score will be incorrect, see Keyword matching is basic for further details.

The plugin in this directory must be added to a script that runs in an open-source ETL package called Kettle. There are three ways of running the scripts:

1. With a user interface, using a Kettle application called 'Spoon'. Spoon also allows you to edit and debug the scripts.
2. From a command line, using a Kettle application called 'Pan'.
3. From another Kettle script using the 'Job' step.

• Steps
  • AMQP Kettle Step Plugin
  • Execute Documents Plugin
  • JSON Decoder Plugin
  • OData Plugin
  • Raise Event plugin
  • Select Documents Plugin
  • Sentiment Score Plugin
    • Sentiment Plugin Licensing
    • Using the Sentiment Score Plugin
    • Sentiment Score Plugin features and limitations
  • Twitter Plugin