1

I'm with a task to deploy "something" capable of doing what Google does, but for PDFs. Could be something like "Google Desktop Search", but ideally the app would run on a server.

Here's some background about my company:

  • we are a government agency. We have hundreds of PDFs with news clippings from sites that mention our agency or some subject of our interest
  • our CEO often gives speech to various communities, so he needs to be briefed on the fly about a specific subject before he goes to a meeting
  • we thought indexing our PDF data would be a good start, but in the long term we aim for a more structured data analysis.

So far I'm evaluating these 2 tools:

thanks!

  • How are your PDFs created exactly? Are they simple scans from hardcopy? Indexing PDFs requires that text within the PDF is represented as ASCII or Unicode strings (in the case of PDFMiner). If some of those PDFs are image scans from a newspaper cutting for example then most likely the text will be represented as an image which isn't indexable. Your problem is then not one of indexing PDFs, but converting your existing PDFs via OCR and then getting some hardware that can scan and OCR at the same time. Fujitsu do some really good desktop scanners that accomplish this - fi7180 for example. They – essexboyracer Apr 17 '17 at 5:22
  • It's still n ot clear what you're asking for. "Doing what Google does" is both vague and of extremely large scope. Do you want to search text within PDFs? Do you want to perform OCR on all your PDFs and keep the results in some external searchable index? Do you want to embed OCRed text within the PDF? etc. Also - government agencies don't have CEOs :-) – einpoklum Dec 14 '17 at 22:55
1

Five years ago I customized Alfresco to perform this exact task.

Alfresco is a document management server, which means you can upload files (including PDFs) to it, and your colleagues can download them or read them online. Each document has metadata (author, date, keywords, etc, you can also add your own metadata fields).

You can integrate Tesseract OCR into Alfresco. Here is the technical procedure: http://www.seedim.com.au/content/alfresco-search-pdf-images-using-transformations-and-tesseract-ocr

Once this is done, you have an OCR "action". The next step is to have this action executed on any PDF that gets uploaded. You can easily do that by creating an Alfresco rule. Now, after someone uploads a PDF containing the word "hello", when people search for "hello" using the Alfresco search field they will see that PDF show up in the results.

Alfresco search PDFs

Both Alfresco and Tesseract are free and open source.

  • this looks really cool! congrats and thanks for the replyl! =) – Lucas Pottersky May 4 '17 at 23:24
0

You could take a look at pythons pdf-miner to extract the information for indexing in a text format. NLTK could also be useful for selecting relevant parts of speech that would be significant.

0

So far my evaluation of Noggle has been positive: https://software4scholars.wordpress.com/2018/08/03/noggle-search-the-content-of-all-your-files/

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.