International Journal of Engineering and Technology

The negative spam reviews are more harmful for hotel services because the impacts of negative information are faster and greater, than positive information. The sentiment analysis for detection of spam review is not effective in today scenario because for making new spam review, the spammers instead of copying exact texts; they are combining two or more text which contains the same sentiment. The spammer used synonyms, morphological words and also shuffled some word to create new negative spam review. In this paper, we have proposed Lexical Chain Based Semantic Similarity (LCBSS) algorithm which gives better accuracy, when compared with the Bag-of-Word (BOW) model and baseline-[1] method. The Proposed LCBSS algorithm has generated feature vector which is used as input to ten supervised algorithm. Amongst ten supervised algorithm, Support Vector Machine (SVM) gave 99.75% accuracy which is the highest accuracy up-till now.