First, how to assign the correct sentiment to a comment when "dictionaries used for automatic analysis of sentiments should be able to take account of: polarity of feeling and the degree of positivity, detection of subjectivity and identification of an opinion, the point of view, the non-factual informations (related to the six universal emotions: anger, disgust, fear, joy, sadness and surprise), syntax and negation"?
Second, how to sort and value the comments that "deserve more interest than others, and [that] can be crucial in the establishment of a new strategy"?
For now, the answer to both questions is: human analysis. Although they have their own interpretation, they apprehend posts subjectively and are slower, human analysts remain today the most efficient in awarding the correct tone to a comment. Tomorrow, technologies as used by Watson, the IBM supercomputer, will able to achieve a degree of precision so that they will become ready to take over the human analysis.
Source : Master Thesis "Mobility sector in Belgium : social media monitoring to built an efficient marketing strategy" 2011, ULB.