The paper presents a method that integrates ontology-based semantic tagging, information extraction and data mining together, automatically learning protein interaction patterns based on a few user seed tuples, and then extracting new tuples from biomedical web documents based, on the discovered patterns. A system, called SPIE (Scalable and Portable Information Extraction), is implemented and tested on the PuBMed to find the chromatin protein–protein interaction and the experimental results indicate the approach is promising.