This research work explores consumers' sentiments through written product reviews. Given a piece of product review, a person likely can tell how much the reviewer likes or dislikes the associated product. This likeness can translate into a numeric scale. For example, reading a product review, one can probably guess how many stars (which is a rating, such as, from 1 through 5) are given by the reviewer to the product. Out of hundreds of words in a particular product review, there might be only few key words that are the key estimators of the reviewer's likeness. This research work explores those estimator type of words, particularly adjectives and other non-adjectives acting as an adjective in its context. By nature, adjectives are words that describe, qualify, modify and give more information about other words. This fact hints an idea of using adjectives in a product review as an estimator for reviewer's likeness. Besides those words defined as adjectives by linguistic definition, this work is also interested in other words that might not strictly be adjectives by linguistic definition but play adjective's role in its corresponding context. This paper reports findings from exploration of the relation between adjectives in a product review and its likeness rating. Experiments are performed on a number of product reviews, book reviews to be specific, from an online bookstore.
