An Information Extraction Method for Digitized Textbooks of Traditional Chinese Medicine

Digital libraries have shouldered the mission of preserving and spreading human culture in the era of information. However, knowledge extraction for digital libraries is not well studied, and that holds back the role promotion of digital libraries from information collector to knowledge provider. This paper presents an ontology-based approach, which extracts detailed attributes of Traditional Chinese Medicine (TCM) from digitized textbooks. According to the characters of digitized textbooks, we propose an extraction ontology that is compatible with both textbook extraction and TCM theory. To improve extraction tolerance for OCR errors, we extract features of different aspects. Finally, a structured pattern based extraction method is adopted to minimize extraction supervision. The result shows that our method is a practical and robust exploration to address the problem of information extraction for digitized textbooks of TCM.