Utilizing the Open Movie Database API for Predicting the Review Class of Movies

In this paper, we present our contribution to the Linked Data Mining Challenge 2015. Our approach predicts the review class of movies using external data from the Open Movie Database API (OMDb-API). We select specific features, such as movie ratings and box office, that are very likely to describe the quality of a movie. With RapidMiner we utilize these features and apply three basic classification algorithms to train and validate the prediction model using a 10-fold cross-validation. The results of our evaluation are interesting in a two-fold way: (i) few movie ratings from professional critics provide a higher accuracy (accuracy ≈ 0.94) than many ratings from users (accuracy ≈ 0.7 ), and (ii) the Decision Tree classifier (accuracy ≈ 0.83) outperforms Naive Bayes (accuracy ≈ 0.73), whereas k-NN is not suitable at all (accuracy ≈ 0.53).