For clustering you will need to identify keywords and clustering will be based on keywords which may not give good results. Instead you can go for opinion mining which is part of text mining.

Thank you Kavita Oza , Opinion mining (Sentiment analysis) will be used as a first step in extracting words but how to define the aspects from the extracted words is my question , do u think clustering is still suitable for doing that ?

Hi Sumaia,

You could extract features from text using a cluster method. Precisely, this can be done using the cluster membership. Here, you need to tokenize the documents first.

Another feature extraction from text also includes the implementation of, e.g., tf-idf method, stemming, etc.

HTH.

Samer Sarsam, PhD.

I think the clustering is not fit for this. In my opinion, the best method for aspect extraction is by using the syntactic-based approaches and/or knowledge-based with relations dependency approaches for extracting explicit and implicit aspects.