| Popis: |
Customer reviews, a.k.a. word-of-mouth reviews, have been important resources of information for text mining. They naturally include both positive and negative opinions on the products or services, as well as neutral observations helpful for everyone who is about to purchase the products or about to decide what to do with the product or the service. Among many customer reviews, we focus on cosmetic reviews and propose a machine learning method to predict the usefulness of an arbitrary review. We impose two conditions on the customer reviews. The first condition is that the review should have certain amount of “Likes” given by anonymous users watching the Web site. The second condition is that the time has sufficiently passed since the review has been posted on the internet, in order for us to make sure the votes by the users are at the stage of conversion. We propose a regression model for predicting the usefulness of customer reviews, introducing twenty two features computed in advance by the training data. We also introduce the BoW (Bag-of-Words) model, having more than 8,000 features, as a baseline method, and conduct the comparative experiments. The results demonstrate that the proposed method outperformed the baseline method in terms of RMSE (Root Mean Squared Error) and R-squared. For future work, we expect our proposed features can be applied to predict the usefulness of other customer reviews. |
