Question about feature-based sentiment analysis

puteri_prameswaputeri_prameswa Member Posts: 3 Contributor I
edited November 2018 in Help
Hi!

Sorry for the newbie question. I'm using RapidMiner for my final thesis regarding feature-based sentiment analysis. Feature-based sentiment analysis is a part of sentiment analysis which seek sentiment (positive, negative, or neutral) in every word (or feature, mostly noun). However, I face difficulties in finding the right tutorial for this task. Can anyone help me with this? I'm running out of time.

Thank you!

Respectfully,
Puteri Prameswari

Answers

  • MartinLiebigMartinLiebig Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 3,503 RM Data Scientist
    Hi Puteri,

    great to have you here on the forums!

    Have a look on this video:
    http://vancouverdata.blogspot.de/2010/11/text-analytics-with-rapidminer-loading.html
    it is a bit old but i think still valid.

    Best,
    Martin
    - Sr. Director Data Solutions, Altair RapidMiner -
    Dortmund, Germany
  • LaurenPlummerLaurenPlummer Member Posts: 4 Contributor I

    Hi Puteri,

     

    The Rosette Text Analysis extension for RapidMiner Studio has an operator called "Entity Sentiment", which will extract entities ((like people, places, products) from your unstructured text in an ExampleSet and return the sentiment of the entities within the submitted text. An update will be released soon to return the sentiment of even more entity "types", like dates, email addresses, etc.

     

    Rosette has a lot of models pre-trained and ready to use, if that's more of the workflow you'd like to use. Just sign up for a free account of the Rosette API and download that extension to RapidMiner and you should be able to pass in unstructured English or Spanish text in an ExampleSet to the "Entity Sentiment" operator, and see every entity and its associated sentiment result (positive, negative, or neutral).

     

    Let me know if you have any questions!

    -Lauren

Sign In or Register to comment.