Evaluating A Customer Reviews Of Products Sold Online

1530 Words7 Pages
Bing and Liu discuss ways to enhance customer satisfaction in this paper by studying customer reviews of products sold online. The paper also describes the problem of generating feature based summaries based on these reviews. According to the paper, a customer review of a particular product is composed of three parts, namely:
Different features of the products that the customers have expressed their opinion,
Identifying positive and negative opinions based on their reviews and finally
Composing a summary based on information obtained.

The main advantage of using feature based summary is that it enables customers for explicit reviews on a particular product. This also enables merchants to study the growth of their products
…show more content…
The system crawls all the information (customer reviews) from the particular product and starts to extract the feature words. The mining classification is divided into three subtasks. First, POS tagging, most product feature in reviews contain nouns and phrases. POS enables to identify these tags words by word. The authors have used NLProcessor linguistic parser to split the sentences and produce POS tag for every individual word. Some pre-processing is performed to remove stop words to avoid big POS tag lists. Fuzzy matching was also used to deal with misspelling and word variants. Second, identifying frequent features, features that are particularly used by customers in many reviews. To put in context, words, or sentences that occur together in many reviews. The authors use association mining to list out these words. An apriori algorithm based association miner is used to identify these features/words. The gathered features cannot possibly be genuine product features. Pruning enables to remove the unwanted features/words that don’t match to product opinion or feature. The author uses two pruning methods and compares their efficiency. While using compactness pruning, it looks for features that is composed of two or more words and identifies if they make
Get Access