Gartner shows taxonomy and ontology management on the wane #taxonomy


In their 2016 Hypecycle for Emerging Technologies enterprise taxonomy and ontology management are headed into the Trough of Disillusionment (the lower saddle of the S-curve). Interesting. 


What will be interesting to see is where it heads next. It could either continue its dive into the abyss; or it could bounce from a yet-to-be-seen bottom and begin its slow and steady rise up into the plateau of productivity. 



Anyone care to provide a backward look at that particular technology's ascent?? Might be helpful to better understand what people were imagining as it hit the Peak of Inflated Expectations; and then discuss whether there is actually going to be a 'there' there or not going forward,



A number of reactions to this report – first, interesting that taxonomy is listed as heading for trough pf disillusion and at the same time it is listed as a component of emerging trends – part of the perceptual smart machine trend


As a recovering taxonomist (now text analyst), I’m not sure that it really went through an inflated peak of expectations, but it does seem to be entering the slough of disillusionment.  It seems to me that the main reasons are:

  1. The difficulty of building an enterprise taxonomy and even more difficulty in maintaining it.

  2. The difficulty of applying the taxonomy to documents

  3. Which leads to severe limits on the applications that the taxonomy can be used for


    #1 – this has been getting easier than ever with growing taxonomy expertise and number of taxonomists -  for example, Taxonomy Boot Camp in DC has seen its numbers grow year after year, so much so, that Info Today is opening a new Taxonomy Boot Camp in London this fall. If you’re in either city this fall, come by – I’m giving talks at both. There are also better text analytics tools – text mining for clusters of co-occurring frequent terms, noun phrase extraction, and simple exploratory categorization rules can all be used to build taxonomies faster and cheaper..  Also, I see evidence of a trend toward smaller, more modular or faceted taxonomies rather than one giant enterprise taxonomy.


    #2 – This has never been overcome when manual approaches are tried. You have to hire too many librarians/metadata specialists to tag thousands or millions of documents or deal with all the known difficulties in using authors to tag their own documents – they won’t do it or do it so badly that the results are worse than no tags.  There is also inter and intra author inconsistency.  The answer I see more often is to use text analytics tools auto-categorization and entity extraction working with authors to semi-automatically tag documents.


    #3 – If you use text analytics, this can also be overcome.  By being able to tag better and cheaper you open up a broad range of applications that taxonomies can be used to support.  And the combination of text analytics and taxonomies is a very good way to add depth and intelligence to those applications – BI, CI, customer relations management, fraud detection, expertise location, KM community collaboration, enterprise social networks, etc., etc.

     In other words, adding text analytics to taxonomies solves (almost) all the problems.

Matt Moore <innotecture@...>


Thanks for the insightful comments Tom R. I would agree that I don't think Taxonomy & Ontology Management ever hit the peak of inflated expectations. I would also note that Gartner puts mainstream adoption of this technology 10+ years away - the same as autonomous vehicles.

Now, I think adoption will happen quicker than that but that the way these tools work will dramatically change over the next few years - exactly due to the developments that you refer to (text analytics, autocategorisation, machine learning).