A Study on Information Extraction from PDF Files

Portable Document Format (PDF) is increasingly being recognized as a common format of electronic documents. The prerequisite to management and indexing of PDF files is to extract information from them. This paper describes an approach for extracting information from PDF files. The key idea is to transform the text information parsed from PDF files into semi-structured information by injecting additional uniform tags. An extensible rule set is built on tags and other knowledge. Guided by the rules, one pattern matching algorithm based on a tree model is applied to obtain the necessary information. A further experiment proved that this method was effective.