Considering the wide spectrum of both practical and research applicability, opinion mining has attracted increased attention in recent years. This article focuses on breaking the domain-dependency barrier which occurs in supervised opinion mining strategies by using a semi-supervised approach, which ensures domain independence. Our work devises a generalized methodology by considering a set of grammar rules for identification of the opinion bearing words. We focus on tuning our method for the best tradeoff between precision and recall and time. Moreover, as the seed words are not specific to a given domain, we claim again that the approach is domain independent.