Hello,
I read some papers they used Pre-processing steps with text that will classify based on Sentiment Analysis.
My question is, can I use text Pre-processing techniques in the sentiment analysis classification, such as the Stop Words Removal and Stemming techniques? If can, it will cause Negations Words or Negation Prefix to be deleted, such as ( I am Not happy), will be (happy) after we use Stop Words Removal technique or (unlucky) will be (lucky) after we do Stemming process. That's mean the sentence that should classified into negative class will be classified into positive class. How to deal with that?
Thanks to answers in advance.
https://towardsdatascience.com/updated-text-preprocessing-techniques-for-sentiment-analysis-549af7fe412a
Article The Role of Text Pre-processing in Sentiment Analysis
Yes, data preprocessing is a first step of data analysis. For a bit details you may look at https://towardsdatascience.com/nlp-text-preprocessing-steps-tools-and-examples-94c91ce5d30 too.
Of course, you can preprocess analyzed texts. However, from my experience it doesn't significantly improve classification results. I've studied it in my PhD thesis and depending on the type of preprocessing or how you weight/discretize feature values classification results can become either slightly better or worse because it is actually data overfitting. I wouldn't remove stopwords since they can be psychologically very helpful (see Pennebaker's Article The Development and Psychometric Properties of LIWC2007
). I would rather stick to choosing a good classifier: it is better to use probabilistic ones such as NaiveBayes if you classify short texts or sparse instances; if you classify long texts I would rather use an analytical one such as SVM. You can check it yourselfThesis Opinion Mining and Lexical Affect Sensing
Sentiment analysis is a NLP problem. Thus, preprocessing will be one of the phases of the Sentiment analysis .
Check this and hope it will be useful for you:
Conference Paper Various Pre-processing Strategies for Domain-Based Sentiment...
hello
in this state preprocesding unuseful
because I am used pre processing and remove stop word and stemming to store words that enable me to easy looking in the search
but in your state the meaning of statement changing
- Badges
- Science topic