I would like to classify binary Arabic sentiment analysis corpus using machine learning techniques.
What can be the features for training and testing data for a sentiment analysis classifier?
Many features can be used the Sentiment Analysis field, as example : word existence, word count, or TF-IDF.
Many Classifier types have been used in the case of Arabic language : SVM, KNN, Naive bayes, etc.
I advice you To take a look at this URL:
http://www.springer.com/cda/content/document/cda_downloaddocument/9783319181165-c2.pdf?SGWID=0-0-45-1516988-p177350424
and
http://www.emnlp2015.org/proceedings/EMNLP/pdf/EMNLP299.pdf
For which authors are proposing a Twitter content dataset at this link :
https://github.com/mahmoudnabil/ASTD
There is also some other resources in this link:
http://saifmohammad.com/WebPages/ArabicSA.html
Good Luck.
Dr. M'hamed MATAOUI
https://github.com/hadyelsahar/large-arabic-sentiment-analysis-resouces
http://www.springer.com/cda/content/document/cda_downloaddocument/9783319181165-c2.pdf?SGWID=0-0-45-1516988-p177350424
https://github.com/mahmoudnabil/ASTD
http://saifmohammad.com/WebPages/ArabicSA.html
Hi Maytham,
For major document classification, the features are bunch of words that should be related to the topic your working on, and also associated with target class. However, the tokenisation technique is quite important to characterize these features during preprocessing stage where you need to look at aspects such as stemming, stopwords list, etc.
HTH.
Samer
Many features can be used the Sentiment Analysis field, as example : word existence, word count, or TF-IDF.
Many Classifier types have been used in the case of Arabic language : SVM, KNN, Naive bayes, etc.
I advice you To take a look at this URL:
http://www.springer.com/cda/content/document/cda_downloaddocument/9783319181165-c2.pdf?SGWID=0-0-45-1516988-p177350424
and
http://www.emnlp2015.org/proceedings/EMNLP/pdf/EMNLP299.pdf
For which authors are proposing a Twitter content dataset at this link :
https://github.com/mahmoudnabil/ASTD
There is also some other resources in this link:
http://saifmohammad.com/WebPages/ArabicSA.html
Good Luck.
Dr. M'hamed MATAOUI
https://github.com/hadyelsahar/large-arabic-sentiment-analysis-resouces
http://www.springer.com/cda/content/document/cda_downloaddocument/9783319181165-c2.pdf?SGWID=0-0-45-1516988-p177350424
https://github.com/mahmoudnabil/ASTD
http://saifmohammad.com/WebPages/ArabicSA.html
Dear Maytham,
These are good links for corpus dedicated to Arabic sentiment analysis:
https://github.com/iamaziz/ar-embeddings
https://github.com/hadyelsahar/large-arabic-sentiment-analysis-resouces
https://github.com/Lab41/sunny-side-up/wiki/Arabic
Good luck.
Best regards.
For more corpora in Arabic in general, please refer to my new survey:
Arabic natural language processing: An overview
https://arxiv.org/abs/1903.02784
- Badges
- Science topic