Extracting information from PDF documents for use in automatic indexing of e-books

Abstract

The number of electronic books that enter libraries in PDF format is greater every day. Complicating and making it almost unfeasible for some processes, traditionally carried out manually by librarians such as the assignment of subjects, to be done. In this context, it is necessary to design and develop applications that assist librarians. Taking this into consideration, we present in this work the evaluation of tools for extracting information from books in PDF format that could be used later as raw material for an automatic indexing system. To do this, we carried out a first evaluation of five software (PDFMiner.six, PDFAct, PDF-extract, PDFExtract, and Grobib), later, as PDFAct achieved the best performance, we did a second evaluation to find out their ability to identify and extract information from the books such as titles, indexes, sections, titles of tables and graphs and bibliographic reference which are relevant information for any indexing system. It is concluded that none of the evaluated tools adequately extracts the different parts of PDF books, although PDFAct has achieved a better performance than the rest.

Keywords
Software evaluation; PDFMiner.six; PDFAct; PDF-extract; PDFExtract; Grobib; Automatic indexing