A knowledge graph is an approach to representing knowledge within some sort of knowledge organization system for some sort of knowledge framework. In his article, Embracing Complexity - Conclusion, Tony Seale explains how to built your organization's knowledge graph.
In summary, Seale believes that an organizational knowledge graph should include:
- Networked data
- Networked cloud
- Networked artificial intelligence
- Unified network that pulls all three of the above together
Then, there are three practical tools that make a knowledge graph go:
- The graph adapter
- The data service
- The graph neural network
Using this approach, a knowledge graph can then give an organization the ability to connect all it's critical data in the entire organization together. This gives both humans and artificial intelligence access to all that data. Seale is not specifying any
specific technical implementation. He seems to be suggesting using a
semantic network and
network theory which is part of
graph theory.
I buy all that Seale is saying but would augment that description a bit.
Sometimes knowledge and knowledge frameworks extend beyond just one organization. For example, the area of knowledge related to financial accounting, financial reporting, auditing of financial statements, and financial analysis spans across pretty much every enterprise and may even be shared by regulators.
A professional oriented knowledge framework; such as accounting, reporting, auditing, and analysis; is a structured approach to providing such a knowledge organization system for a widely shared area of knowledge. A knowledge framework relates to how the knowledge within an area of knowledge is created, shared, and evaluated. Such professional knowledge frameworks need governance mechanisms. Governance helps manage epistemic risk.
The graphic below describes how a knowledge framework works and forms a virtuous cycle (a.k.a. feedback loop or causal chain) when it works well. The objective is to achieve a mutual agreement and understanding for the entire area of knowledge which is maintained over time.
As pointed out by Inhabiting Babel, meaning is not something inherent in just words or objects. Meaning is consciously and deliberately produced through shared human systems, shaped by context, and sustained by communities of stakeholders.
Machines interpret information based on predefined rules. Humans understand information based on context, skills, experience, research, observation, and reasoning. You cannot automate understanding.
Understanding is when something makes sense to you well enough that you can use it, explain it, or see how it fits into the context with or connects to other things you know.
Meaning is the content. Understanding is the competence (skill and experience).
Meaning is an encoding. Understanding is an active process: interpreting, integrating, contextualizing, applying, reasoning.
Meaning is the map. Understanding is the ability to navigate the terrain.
Meaning is
intersubjective; an agreement between parties. Understanding is individual.
Meaning enables understanding. Understanding validates meaning. There is a feedback loop between meaning and understanding. Without stable meaning, understanding collapses into ambiguity. Without understanding, meaning is inert, no uptake.
A knowledge framework helps professionals within an area of knowledge create, maintain, and utilize a knowledge organization system. Subject matter experts create and maintain the elements, connections, conditions, and facts within their subject matter. To do this a
logical theory is used to explain and discuss how the knowledge in the knowledge organization system works.
The knowledge in a knowledge organization system tends to not be used only as individual pieces of knowledge or used all at once. Fragments of knowledge tend to be used, fragments of the full knowledge graph. Structures are used to describe those knowledge graph fragments.
Atomic Design Methodology can be used to help subject matter experts think about their structures. These "
digital information organisms" are how subject matter experts interact with their knowledge. This layer hides the technical complexity of the knowledge graph, exposing meaning and enabling understanding by the professionals with the skills and experience in an area of knowledge.
One example from accounting, reporting, auditing, and analysis is the
general purpose financial statement. Other examples are accounting working papers, audit working papers, financial analysis models.
Another thing that is necessary for a professional such as an accountant, auditor, or analyst to interact with a knowledge framework or knowledge organization system is user tools. Not technical tools, user tools specifically designed to work with the structures of a specific area of knowledge.
My point is that
structures and
models make a knowledge framework and knowledge organization systems "go" for business professionals that need the system to be both powerful and easy to use. Atomic Design Methodology helps one think about such structures and models.
This screenshot below is an example of networked data, networked cloud, networked artificial intelligence, and a unified network to pull everything together:
The graph adapter, the data service, and the graph neural network need to interact with the networked data, networked cloud, networked artificial intelligence, and the unified network which pulls everything together in a manner that is approachable to business professionals.
Graph theory is great; but design is necessary also.
Oh, and then there is the problem of bad data and bad systems that need to be fixed before you build a nice shinny digital process. I will leave that for another time.
Additional Information:
Comments
Post a Comment