we need to analyze Italian facebook text data for identifying users expressing positive or negative 'sentiment' about the service level of a specific company.
Do you have any suggestion or examples on how to build such a workflow for Italian text interpretation?
Any suggestions on how do it in a smart way, avoinding tons of IF ?
Thanks a lot in advance!
if you have labeled data you can build a classifier. Here is a blog post including data and workflow that describes how to do this using KNIME: https://www.knime.org/blog/sentiment-analysis
If you don't have labeled data you need a dictionary with italian positive and negative words. Count the pos and neg words in each document and compute a sentiment score on these numbers. Unfortunately I did not find any italian subjectivity corpus or sentiment dictionary so far. For English the MPQA corpus is suitable: http://mpqa.cs.pitt.edu/ . If you find an italian dictionary, please let me know.