A RapidMiner framework for protein interaction extraction

During the last 10 years researchers have proposed many approaches to automatically extract protein-protein interactions (PPIs) from scientific papers. However, the lack of a unified implementation and evaluation framework complicates the development of new PPI methods and makes the comparison of existing methods difficult. In this paper we present such a framework that is built as an extension of RapidMiner. Next to providing a platform for evaluating and comparing different text mining methods for PPI extraction, the framework can also be leveraged to build a standalone application that can accumulate the mined interaction data and make it available to biologists for querying. We illustrate the utility of the developed framework with a prototype of a protein interaction search engine.