Re: Model of an intelligent knowledge extraction in organizations #extraction #definition

Chris Collison


I’ve always been drawn to the way Larry Prusak describes information – as ‘a message with a sender, a receiver and an intent to inform’.  I’d happily include within Larry’s definition, any messages explicitly encoded chemically as ant trails, digitally as content, or through the medium of dance by a waggling honey bee )


I think this is one of those areas which we’ll all take different stances on, and all tailor where and whether we draw the boundary line – or overlap zone -  for the needs of specific clients.   It’s (as Nick M once wrote) probably a bit of a cul-de-sac conversation for KM enthusiasts – but it’s been really interesting to read the different perspectives in the safe-space which is SIKM! 



From: <> on behalf of Stephen Bounds <km@...>
Reply to: "" <>
Date: Saturday, 16 January 2021 at 12:12
To: "" <>
Subject: Re: [SIKM] Model of an intelligent knowledge extraction in organizations


Hi Chris & all,

I am not a fan of the tacit/explicit dualism because I think it is too reductive and privileges a human-centric view of knowledge. I agree with Nick that it is perfectly fine for something to be both information and knowledge.

Here are just a few scenarios which I think challenge traditional views of tacit vs explicit, and explicit knowledge = information:

  • Ant scent trails provide an active, contextual history and guide for the behaviour of the hive. By any reasonable metric this is a form of "knowledge" despite being entirely "external" to the ants.
  • Mandatory procedural instructions (PIs) very distinctively corral the efforts and choices of an organisation. From the perspective of an outsider, while the emergent problem solving rhythms and algorithms are observable, the organisational knowledge encoded through multiple overlapping documents and put into practice by staff is quite opaque to the client and definitely not directly extractable as knowledge.
  • The contents of the text messages on my phone are parsed and actioned as if they were a form of external memory, quite distinct from the information processing involved in reading a newspaper article or book. (For example, think about the rich visual, audible, and emotional memories that that may be triggered from the act of reading a text.)

Rather than tacit vs explicit, I believe that the key transition occurs as we cross a system threshold. Inside the system threshold, it is meaningful to talk about its knowledge; outside, we must talk about transmitting information. New information can only be accepted as knowledge into that system once it achieves a certain trust threshold.

Thus, a written process sent through by head office represents information received by a staff member, but can be knowledge once incorporated into the execution of their role's systemic practices.

The switch in language between a role and a person is important since it represents an enlargement of system scope; employees are often asked to undertake a role "performance" which includes a broad scaffolding of policies, processes, technology, and person-to-person relationships.


Stephen Bounds
Executive, Information Management
E: stephen.bounds@...
M: 0401 829 096

On 16/01/2021 6:02 am, Chris Collison wrote:

I’m with you Fred.

...and from my perspective (which not everyone shares!), I would add that explicit knowledge = information.




From: <> on behalf of bill@... <bill@...>
Sent: Friday, January 15, 2021 7:57:21 PM
To: <>
Subject: Re: [SIKM] Model of an intelligent knowledge extraction in organizations


Hi Fred


I say the following knowing the risk with which I am placing myself 😊…never been a fan of “implicit” and so I am not a fan of Polanyi’s definition.  My point, based on years of practical consulting application, is that “tacit” knowledge can be elicited, harvested, captured and distilled to make it searchable, findable, accessible and usable/reusable. There is a skill and craft involved IMO. My definition of knowledge is based on the concept that it consists of information and experience = knowledge. For practical, not academic or theoretical application, this simple, practical definition I find has meaning in a business or operational environment.  This has been my experience.






From: <> On Behalf Of Fred Nickols via
Sent: Friday, January 15, 2021 11:22
Subject: Re: [SIKM] Model of an intelligent knowledge extraction in organizations


FWIW, in relation to Bill’s use of “explicit” and “tacit,” I find a third category useful; namely, “implicit.”  I define “implicit” as knowledge which hasn’t been articulated but can be.  Tacit, by Polanyi’s definition, can’t be articulated.  And explicit, of course is knowledge that has been articulated.


When it comes to capturing implicit knowledge, lots of folks have been doing that for many years.  You will find them in the training and human performance community.


For more, see a piece I wrote for the KM Yearbook (2000-2001).





Fred Nickols, Consultant


My Objective is to Help You Achieve Yours





From: <> On Behalf Of bill@...
Sent: Friday, January 15, 2021 10:41 AM
Subject: Re: [SIKM] Model of an intelligent knowledge extraction in organizations


Dear Colleagues


Been enjoying this timeless discussion on K capture and the extended inference about tacit knowledge and the subtleties of human language in the sharing in the context of the below thread.


Believe it is generally recognized that technology cannot yet get what is one person’s mind and transfer it to another person’s mind. At the heart of the issue, I believe, is the skill set that is required to capture or better said, to elicit what someone knows or can share, to distill and make sense of that knowledge whether it be explicit or tacit, and then to characterize it for reuse within an organization (fit for purpose and context relevant) and then to make is searchable, findable, downloadable, and usable/reusable.

I believe that this is a reasonable approach to as stated below “…to study KM processes, identify patterns and see if technology can address that in a narrow sense.” 








Learn more about the solutions and value we provide at





From: <> On Behalf Of Sam Yip via
Sent: Thursday, January 14, 2021 21:41
Subject: Re: [SIKM] Model of an intelligent knowledge extraction in organizations


We are still very far from seeing any AI technology that can intelligently extract human knowledge in a broad sense. A main challenge is that of data, and in particular textual data.  Data has to be structured in a certain format to be consumed by an AI machine, but in the case of textual data, the number of permutation and subtleties in human language makes this task tremendously difficult. But perhaps the biggest challenge is the limitation of computation power. The most powerful computer today has roughly 1 billion neural connections/synapses, similar to the amount of synapses that a honey bee has vs 100-150 trillion synapses of a human brain. It has been suggested by computer scientists that we are actually just approaching bee-level intelligence, and we are not even there yet. 

That said, AI should be sophisticated enough for a focussed application, especially if it’s built around narrow use cases and trained for specific domains. e.g. the technology to recognise and extract names/entities from homogenous, unstructured text is fairly advanced, and so is the ability for a machine to auto-index/classify documents and textual data. Information extracted as such, combined with data visualisation techniques, will go a long way in enhancing the comprehension of what’s contained in an organization's textual data, and that is a possible KM angle you can look into. 

Instead of looking at broad ways “to extract automatically all types of knowledge in an organization” (I don't think this exists), it might be worthwhile to study KM processes, identify patterns and see if technology can address that in a narrow sense. 

Join to automatically receive all group messages.