Dear Mathieu,  Pascal;


We would like to thank you for accepting our manuscript subject to the 
recommendations of the reviewers.  We will like to thanks the three reviewers whose comments, we believe,  have allow us to improve the second version of the paper. We have made many modifications to the paper in addition to those suggested by the reviewers.

Many thanks.



* Reviewer # 1 


** relation between the presented work and the application (business entity, time, etc.):
The presented work on OM is part of an ongoing project and a particular application (in its final phase) in that project. In the paper we mainly focus on the opinion mining work. We clarify this fact in the paper.


** originality of the work:  
We present two approaches one which relies on simple n-gram based features (these have been used in previous work), the second approach is completely new, consists on the identification and use of semantic-based features derived from 
an existing resource, and on the particular interpretation we give to the lexical resource (no need for word sense disambiguation for example). We believe this is clear now in the conclusion.

** state of the art:
We have expanded the state of the art considerably and established relations between our work and similar work in the field.

** choice of ML approach:
We have justified our choice of ML (SVM) 
by relying on past work on OM and other areas in human language technology. 
We have also tested other ML approaches (Weka machine), but have obtained better results with SVMs. We clarify all this in the paper.


** GATE:
We have included more information on GATE.

** screen dumps:
We have improved the figures.

** contradiction introduction and experimental section:
We have specified what features we investigate in this approach.

** parameter estimation:
We have followed (Li&al'09) specification on how to set the parameters of the SVMs. This is explained in the paper.


** summary:
We have edited   the summary which we hope better reflects the content of the paper.




* Reviewer # 2

** user of the system.
It is true that we have presented a mainly quantitative evaluation. But in the project MUSING qualitative evaluation is being carried out (we state this in the conclusion). Note however that because this is an on-going project full user evaluation is not completed yet.
** integration
We have concentrated in the paper mainly on the OM aspects of the work.
Other aspects such as integration have been deal with in other papers (which we cite in the paper) and will be described in other works.
** typos:
We have corrected typos and remove some problematic phrases.

* Reviewer # 3

** data sets
Although the data sets seem smaller than other data sets used, the results 
we obtain are comparable or better than previous work. 

** article organization
We have re-organized the paper, in particular many subsections in the experimental part of the paper have been re-organized.