Tagging and Taxonomy - any advice (part two)?

Replies to This Discussion

Permalink Reply by John Gregory on December 8, 2008 at 8:53am

There are about 70 vendors out there, I've attached a spreadsheet I put together for a U.S. government contract; it is just bare bones. The big divide in vendors is whether the software is linguistic based and whether it requires "training sets" of documents. In my experience, I want some linguistic element and training sets are a lot of trouble.

Attachments:

• Taxonomy Comparisons.xls, 52 KB

▶ Reply to This

Permalink Reply by David Warren on December 8, 2008 at 9:54am

Thank you very much, this is very helpful!

▶ Reply to This

Permalink Reply by Robin Sherman on January 8, 2009 at 2:02pm

David, how are you coming along with this project?

Robin Sherman
Editorial & Design Services

For Books, Journals, Magazines, Manuals, Newsletters, Internet

— Information Architecture, Content Development, Organization, Improvement
— Developmental, Substantive, Copy Editing
— Publication Critiques
— Publication Design
— Typesetting, Typography, Layout
— Speaker, Seminars, Workshops

Small Publisher and Non-Profit Specialist

▶ Reply to This

Permalink Reply by OlegR on February 5, 2009 at 11:59am

Hi David,

I work at Nstein, so my 5 cents might be a little bit biased and you also might have heard what I'm going to say already. In any case, having this message in the public could be helpful for someone else.

There are 2 types of text mining systems out there - hosted/perpetual license and SaaS. (Nstein, Temis is license based, Inform is SaaS).

The main difference is owning the metadata. While having to pay a smaller fee out of your operational budget might sound very seductive, in the end, when going with SaaS you do not control your metadata, your taxonomy, your dictionaries, etc.

It's like maintaining your custom made, heavily modded and beloved million dollar car in a "Joe Shmoe" car shop. The point of text mining is to fine tune access to your content both on your front and back end. When you don't control your taxonomy, entity lists, dictionaries and you don't have the meta-data in house - you are just tapping the top of the top of the iceberg of possibilities, while paying more over a 3 year period of SaaS subscription than you would pay for owning a licensed and fully controlled text mining engine.

Second point I wanted to make is importance of owning, and dynamically controlling your taxonomy (-ies). Who knows what's going to happen in your industry/area. Imagine being able to suddenly adjust and deepen your taxonomy for a newspaper in Virginia area just when the shooting happened. How much more traffic could you retain on your site?
Or imagine having one general taxonomy for your overall content, and micro-taxonomies to answer special projects, micro-sites, etc. Or having a separate taxonomy for every of your syndication partners, thus providing them with enriched content (chewed and digested for them) - that could greatly increase your syndication revenues.

Thirdly - relevant content linking. When your text mining engine does not have access live to your DAM or other repositories, it cannot provide you with relevant content links between your regular content, your UGC, your partners' content, etc.
What's the point? One of the most successful features of text mining is all those hundreds of relevant links and inline-tags that that keep that user on your site for 30 minutes, instead of 30 seconds.

Below I'm attaching a visual example of all the features, properly installed and configured text mining engine can bring to your side, to your revenues, to your bottom and top line in a completely automated fashion.

▶ Reply to This