Playscript Classification and Automatic Wikipedia Play Articles Generation

Abstract

In this work, we aim to create Wikipedia pages on plays automatically by extracting relevant information from various web sources. Our approach involves building an efficient classifier that can classify web documents as play scripts. From the set of correctly classified instances of play scripts, we extract relevant play-related information from the documents and use it to obtain additional information from various sources on the web. This information is aggregated and human-readable Wikipedia pages are created using a bot. The results of our experiments show that classifiers trained by combining our designed features along with "bag-of-words" (bow) features outperform classifiers trained using only bow features. Our approach further shows that good quality human-readable pages can be created using our bot. Such automatic page generation process can eventually ensure a more complete Wikipedia.