Intelligent Wrapping from PDF Documents

Wrapping is the process of navigating a data source, semiautomatically extracting data and transforming it into a form suitable for data processing applications. The semi-structured form of web pages, coupled with the availability of business-relevant data, has led to the availability of several established products on the market for wrapping data from the Web. One such approach is the Lixto methodology [1], a result of research performed at DBAI. Many commercial applications also require the extraction of data from PDF documents. There appear to be no general-purpose approaches to fulfil this need and, as the PDF format is unstructured, this is a challenging task. We are investigating PDF data extraction in the NEXTWRAP project. This paper presents our work in progress, with particular reference to low-level segmentation algorithms.