Workflow with example of Emoticon analysis

Hello everyone,

I am looking for a workflow with emoticon analysis to classify text in negative / positive category based on emoticons in combination with other learners / predictors. May be you have already encountered such a case or have an example workflow. It would be great if you can point me to an example or a resource on this matter. Thank you very much!

Hi,

so far I have not done emoticon analysis and have no example workflow, sorry for that. There are two blog articles about sentiment analsis but they don't cover emoticons:

https://www.knime.org/blog/sentiment-analysis

https://www.knime.org/blog/sentiment-analysis-with-n-grams

Cheers, Kilian

Thank you Kilian, I appreciate it. I have looked into the 2 links you mentioned before posting my question

The emoticon analysis is only a part in my sentiment analysis project with different techniques. 

Hi,

For sentiment analysis using N gram,from where can I get the 1-2 gram positive and negative dictionary.

Thanks,

 

 

 

 

Hi Payal,

the Textprocessing extension does not provides dictionaries. If you have labeled data you can create feature vectors from 1 and 2 (and n) grams and build predictive model based on these features vectors. If you want to use dictionaries, you need to bring your own.

Cheers, Kilian

Hi @kilian.thiel

Can you by any chance provide an example workflow for emoticon analysis these days? :smile:

Best,
Theresa

See if sentimental analysis is close to what you looking for
https://www.knime.com/blog/sentiment-analysis

Hi @reth19so and welcome to the forum -

We don’t have a workflow that I know of for dealing with emoticons in this context, but it’s a good and useful idea.

Apparently (thanks @julian.bunzel!) the StanfordPTB tokenizer can parse emoticons, so in theory they could be labeled using the Dictionary Tagger - but I confess I have not tried this myself yet…

Hi @kuldokk,

Depending on your comfort level you should check this out: https://github.com/cjhutto/vaderSentiment.

I use the Python library in KNIME for sentiment analysis with this tool which will analyze emoticons plus it understands phrases like LOL etc. You can easily tweak it under the hood as well.

You can tinker with how it’s evaluating for sentiment whether whole text or a composite based on sentence tokenizaton using NLTK for Python. They include examples to follow and I’m happy to share how I have my Python nodes setup for this.

There is also an R package out there as well.

2 Likes

Alright I’ll look into that. Thanks @ScottF!