#1

i am trying to do the following scenario
1- user enters a twitter hashtag(trending mostly)
2- sentiment analysis to output polarity detection

i have an idea about how to do sentiment analysis but
sometimes there are opinions about things that don’t relate to the hashtag.
so i want to do semantic analysis to make sure the model understand the context and only output sentences that contains opinions about the hashtag to use them in my sentiment analysis but i don’t know how so i started with learning more NLP i learned all these and don’t know which to use to do what exactly

1-tokenization
2-sentence segmentation
3-part of speech(POS)
4-stemming
5-NER
6-stop words
7-Matchers
8-syntactic
9-bag of words(BOW)
10-TF-IDF

also studied naive bayes classification

#2

Maybe you should take a look at the annotations:
https://developer.twitter.com/en/docs/twitter-api/annotations/overview

Stop words/POS/stemming/matchers/BOW… depends on the correctness of the written words, and the corpus generated from your twitter sample is going to have many misspellings to overcome computationally (lehvenstein distance and other similar procedures).

Also, take a look at https://developer.twitter.com/en/docs/tutorials/how-to-analyze-the-sentiment-of-your-own-tweets. Although it involves Azure services, it can shed some light in your workflow.

Hope helps!

1 Like
#3

what is the difference between annotations and NER ?
user misspelling is not my concern actually someone else should pass the text through grammar checker to correct those
the second link Azure is doing the sentiment analysis I would love to see how they do so I only learned how text blob does it and it was easy to follow and understand

here is a review about iPhone : “my kid is special ,kids are great , iPhone is good screen is very clear
I hate the way (it) handles w/e”
I want a way to make my program understand that kids are great is not part of the review and (it) in the last sentence is part of the review (syntactic analysis for that I suppose)
can’t just do sentiment on this and get great as positive when it has nothing to do with the iPhone
it feels like a classification weather the sentence relates to the topic or not
I am so lost thank you for helping me