Document Embedding

[@olaali4068]

I'm working in a research to analyze Google play users reviews for mobile banking applications.. and the process is splitting data to negative or positive depend on rating + text preprocess + LDA (directly without creating bag of words).. please advise are my steps is right? and if i want one of orange experts to revise my analysis is that available?

[@OrangeDataMining]

Hi! First, LDA requires bag of words to function. So you will need it anyway. Second, LDA can work as a dimensionality reduction technique, but if it is not super necessary, I'd avoid it. Use BOW to end up with an explainable classifier. For consulting see revelo.ai/en.

[@cfzornada]

Hi, Is there a module/script in python to interact with ollama? Thanks from Argentina

[@OrangeDataMining]

GPT is available in the Prototypes add-on. Ollama unfortunately not yet.

[@momenelnasharty1992]

Is this support Arabic language?

[@OrangeDataMining]

Yes. SBERT is multilingual (including Arabic), while there's an Arabic model for FastText.