What's the deal - Identifying online bargains

The Internet is home to an ever increasing array of products and services available to the general consumer. This trend has given rise to a unique category of internet search where bargain seekers have conjugated towards deal collection databases. This is caused, in part, because traditional internet search engines do not perform well in this domain. Unfortunately, these deal databases are costly to maintain due to the heavy reliance on human participation in order to populate them. This has lead to an interest in the development of this class of internet search. Our research focuses on leveraging machine learning and natural language processing to develop a semi-supervised Web page classifier specific to this problem. We describe the design of our classifier with respect to the machine learning model chosen and the training features selected. We compare our model's effectiveness in classifying deal versus non-deal Web pages against other popular machine learning models such as decision tree, support vector machines, and neural net. Our results show that our proposed model performed the best given the features that were extracted for model training and testing.