Open Source Document Management Systems?

Lancer asks: "I sat in on a product demonstration this morning for a Document Management solution my employer is considering. The product looked very nice, lots of useful features for creating a paperless office. Then the hammer came down - $40,000 to get started, of that less than $10,000 would be actual hardware - ouch! So, my question is, are there any free solutions available for document storage and retrieval? If you're not familiar with the concept, these systems generally involve scanning a document (then throwing it away, hopefully), running it through OCR, and putting it into some sort of database (these vary wildly in capability from vendor to vendor), and being able to query the documents out later." Actually, a few don't even bother to OCR, and just store raw scans in the database. Surely something can be built from existing open sourced tools that can do this for less than half this price!

Re:What a great topic

[Pauly]

I would observe then that this company may not be ready for OSS for a number of reasons:

-OSS is not really a direct competitor to commercial software. OSS benefits only those companies who realize the benefit of a long-time investment in their own IT infrastructure and are willing to bear the seemingly high startup costs of time and money.

-OSS is not for companies that trust software more than well-paid, qualified, professional IT staff. I rarely count on just software to handle things like redundancy, backups and version control. That's what trusted system admins are for.

OSS suits companies who realize the benefit of investing money in qualified IT people over expensive software. This should be particularly important to a company that wants to replace the low-tech stability of paper-based systems with the high-tech unknown of a "paperless" system. Paper will never let you down, can you say the same of your software?

-The quality of open source software is directly related to the number of people using it. Even if there were a comparable, open-source rival to the commercial offerings, it's not likely it would be of the caliber of Apache for instance. They'd have to fix/add to the source eventually.

It's a long-shot, but I would recommend that this company develop this application as an open-source project. They stand to augment their own efforts with the assistance of the open-source community.

I reiterate that this is not a overly-complex problem. There is a myriad of successful client-server database applications out there to mimic in an undertaking like this.

Re:What a great topic

[joshwa]

OK, but what about smaller companies who can't afford the time/cost of rolling their own DMS? The commercial solutions have all the caveats you mentioned above, including price, but they do have the robustness that a quicky-job won't have (redundancy, backup, version control, permissions, interfacing with billing software, etc.) Notice I say robustness, not reliability. ;) A an OSS solution would be ideal (reliability-wise), but the points raised in the discussion of OSS RDBMS unfortunately are true here also: it's not enough of an itch. The interest (and the features) just won't be there.

Linux OCR?

[Booker]

While I have seen some OCR projects for Linux (one turns up on freshmeat...) I'm not sure they're robust enough yet.

However, I'd love to see a perl/gtk script (or something) that lets me choose a date, choose a category (bank, auto, mortgage...), scan in a couple pieces of paper, and then packet-write it to a CD... not very full featured, certainly, but it would help get some of the piles of crap off my desk...

---

What a great topic

[Pauly]

I used to work for such a document management company. If ever there was vertical market that has to go away, it's this one. Vertical market software routinely works great as long as you change the way you do your thing and do it "their way." Ask Volkswagen about SAP....

Before you punt that $40,000 into the wind, think of what they are providing:
-Scanning software that creates images for
-OCR/ICR that helps (often incorrectly) capture data for
-Indexing so the images/data can be stored in a...
-Database for future...
-Retrieval and Viewing by client software

These tasks are trivial! With the exception of the OCR portion (which you seem indifferent towards anyhow), these tasks are half done for you. Build your own system and your employer will have a system that suits their needs and is totally flexible.

You'll save money too. Companies like this always have high-cost support contracts, "click" or per-transaction charges, annoying and expensive licensing arrangements, proprietary and closed systems and hit you with consultation fees when you need someone to come to your office to make the damn thing work. Do I sound bitter? I am ....

Build your own system thusly:
Buy high quality server hardware and a consultant if needed. Buy high quality scanners from Fujitsu, Bell + Howell or Panasonic. Shop around for a distributor in your area with a good support program. Think availability of parts and down-time guarantees.

Use the scanning software that comes with scanner. All high quality scanners come with ISIS [pixtran.com] drivers included. Yes, it's windows only but they work really well and have a well documented API. Twain SUCKS for high-volume work like this.

Store the images in a database or on a network drive.

Capture only the data you really need. Anything else is a waste of time and resources. OCR/ICR the image files with TextBridge or ProOCR. They're better than ANY ocr I've seen from these vendors and cost $100 a seat at most.

Store the captured data with the images in the database.

Build a web application with apache/perl that searches through the database for user-input information and returns the image(s) and data to the web-browser already installed on your users' computers.

There are some issues between these steps that need to be glossed over with small applications that are trivial in Delphi or VB. It's so easy, it's a sin to pay any of those vendors for something that I promise you will never truly suit your needs.

Vertical Markets

[Matts]

This is yet another vertical market that Open Source has yet to really break into. And why should it? It's hardly an easy itch to scratch - the problems these companies have been working to solve they've been working on for years.

See also the "Ask Slashdot" about enterprise level RDBMS's.