PDF Metadata Optimization

1
PDF Metadata Optimization

• Masters Project Presentation
• Advisor Dr. Stewart Shen
• Student Roopa Kadiri
• Dec 08, 2006

2
Overview

• Introduction
• Process Overview
• PDF Metadata Update Process
• PDF Metadata and Keyword Search
• Conclusion

3
Introduction

• SEO - Search Engine Optimization
• Search Engine Optimization (SEO) is the practice
  of manipulating aspects of a Web site to improve
  its ranking in search engines.
• PDF Metadata Optimization deals with updating the
  PDFs with proper metadata.

4
Technical Challenges

• Automatically assign Metadata for each PDF
  (Title, Author, Keywords).
• Process PDF and analyze the text inside to
  identify potential candidates for Title, Author
  and Keywords.
• To accomplish the above challenge there is a need
  to choose appropriate Platform, Toolkits and
  develop relevant Scripts.

5
Process Overview

• PDF Metadata Update Process
• Upload PDF document
• Uncompress PDF
• Analyzing and Parsing
• Identifying Title and Author
• Identifying Keyword
• Updating PDF Metadata
• PDF Metadata and Keyword Search
• PDF Metadata Search
• PDF Text Keyword Search

6
PDF Metadata Update Process Diagram
7
Upload PDF Screenshot
8
Uploaded PDF Document Screenshot
9
Script Screenshot (Uncompress PDF)
10
Uncompress PDF

• Uncompress PDF using pdftk tool.
• pdftk is a tool used to manipulate PDF files.
• Here it is used to uncompress a PDF into a
  temporary file which can be parsed and analyzed
  fairly easily.

11
Analyzing and Parsing

• Developed a perl script which parses through the
  uncompressed pdfs, and identifies the text inside
  the pattern ()Tj and groups them based on their
  font value which is specified using Tf values.

12
Uncompressed PDF
13
Identifying Title and Author

• Text streams collected in a list are grouped
  based on their font values.
• Sorted the list to get the first two text streams
  which have the largest font values.
• Above obtained text streams are good candidates
  for Title and Author respectively.

14
Identifying Keyword

• Developed script to parse and look for Keywords
  text in PDF document.
• If found, use the words preceding the Keywords
  text as a candidate for the metadata field
  Keywords.
• If not, as an alternative, find the frequency of
  repeated words (after filtering the stop words)
  from the PDF Text and use top 15 words from them
  as Keywords for the metadata.

15
Updating PDF Metadata

• PDF API2 is the perl module used to assign the
  Metadata for PDF files.
• Used this module to update PDF files with the
  above identified Title, Author and Keywords.

16
Updated PDF Document Screenshot
17
PDF Screenshot
18
Updated PDF Document Properties Screenshot
19
PDF Metadata Search

• Boolean search on Title, Author and Keywords.
• Used PDFAPI to extract metadata from PDF and
  search for the user entered Title, Author and
  Keywords.
• Display the matched PDF documents.

20
PDF Metadata Search Screenshot
21
PDF Metadata Search Results Screenshot
22
PDF Text Keyword Search

• Used xpdf tool to extract text from PDF.
• Loop through the contents of each pdf's text and
  look for the occurrences of the user entered
  Keyword.

23
PDF Text Keyword Search Screenshot
24
PDF Text Keyword Search Results Screenshot
25
Conclusion

• Accomplished automatically updating PDF with
  Title, Author and Keywords so that the search
  engines can pickup the updated metadata while
  crawling.
• Additionally, I have demonstrated metadata and
  full text keyword search of PDF document.

26
References

• Active State Perl 5.8
• http//www.activestate.com/Products/ActivePerl/?t
  n1
• Pdftk tool
• http//www.accesspdf.com/pdftk/
• XPDF
• http//www.foolabs.com/xpdf/
• PDFAPI2
• http//search.cpan.org/dist/PDF-API2/

27

• Thank You