Aspect extraction from online product reviews

Nowadays, online reviews are of great value to a customer, since they can indicate how valuable a product can be. To help a customer gain insight on the information in the reviews it would be beneficial to train a model that can extract aspects of the reviews automatically. In this way, it can quickly be seen if a product is of good quality. The same goes for companies, who quickly want to see how their products are being valued. That is why research needs to be done on the problems that arise when creating such automated models. The contribution of this paper is that it has shown that even though creating a supervised model is more time-consuming than an unsupervised model, the results in the end are worth the time that it takes to annotate data. This is because the unsupervised model has shown to be way worse at mining aspects from laptop reviews.