Auto-Taxonomy and Categorization As A Service (ATaCAAS)

Skip to end of metadata
Go to start of metadata

Motivation

Taxonomy strategies hold great promises, and we nowadays understand that we need some kind of meta-data strategy to automatically and dynamically (loose coupling) connect bits of information. Today´s taxonomy strategies have struggled hard to provide positive return of investment (ROI). Projects experience frequently that their taxonomy changes and they have to keep re-investing in their taxonomy and categorization.

We believe that since the internet search engines started to rule the world, we need to rethink they way we look at information and categorization. With the new and up-coming auto-categorization services, we see a potential of shifting the effort from people to machines in taxonomy and categorization in the same way that the serach engines shifted the information gathering from catalogs to search.

CASE/Implementation strategy

"Pimp my CustomerService"-style service extension, where we use free auto-taxonomy services and object-graphs to generate web 2.0 social-network graphs/presentations of customer clusters to enable pin-point direct mailings to specific interest-areas of the customer clusters.

Value proposition

  • Low initial investment (less than a months effort)
  • Auto-learning/auto-updating taxonomy (we get the new features from the auto taxonomy services for free)
  • No extra maintenance and product cost


Work in progress
You get the idea..

Pimp my service implementation details

  • Only use free and open services and Open Source

Implementation

Labels:
presentation presentation Delete
Enter labels to add to this page:
Please wait 
Looking for a label? Just start typing.

5 Comments

comments.show.hide

  1. Oct 10, 2008

    Mads Nissen

    An important discussion in this area is whether to use a pull/indexing/crawling strategy or rely more on a pub/sub eda service architecture (Norås-style).

    1. Oct 11, 2008

      Thor Henning Hetland

      This is an important part of the exploration of this strategy. Initial strategy will be something like:

      • workers (pull) agents for crawling customer data (web crawling internet company news and such)
      • event-interface (subscribe) to receive categorized documents and taxonomies to be added to the graph and persisted
      • pull mechanism to get the graph

      Does this make sense?

  2. Nov 25, 2008

    Thor Henning Hetland

    http://www.avc.com/a_vc/2008/11/making-twitter.html

  3. Dec 09, 2008

    baardl

    Any interest in discussing this topic at JavaOne or CommunityOne?

  4. Jan 20, 2009

    Thor Henning Hetland

    Interesting MasterData Visualization