A Novel Use For Freeplane

A Novel Use For Freeplane

Postby tylerreed » Tue Mar 25, 2014 1:20 pm

We're using Freeplane to construct quantitative mappings between federal guidance and real-world enterprise architectures. Working in the Health and Human Services (HHS) space, there is federal guidance that describes process maturity levels for State HHS agencies. However, the guidance is vague, highly context-sensitive and sometimes contradictory. Using Freelane, we construct extensive mapping from interpretations of the guidance (that we actually parse using TheBrain). These mappings reflect the quantitative nature of the guidance and ALSO describe calculation schemes that contain references to enterprise architecture elements (processes, roles, systems, etc) that are related to the words. We then built an analytic engine that can read the mappings (we call them quantitative perspectives) and calculate results using values and relationships defined in the architectural models. Architectural models are developed using a standard EA tool.

The result: we can quantitatively show how moving from current-state enterprise to some target-state (say by adopting a new software system or changing a process) will effect various aspects of enterprise maturity from a federal perspective.

Freeplane rocks!
tylerreed
 
Posts: 3
Joined: Tue Mar 26, 2013 2:30 pm

Re: A Novel Use For Freeplane

Postby boercher » Wed Mar 26, 2014 11:34 pm

Hi tylerreed,

thank you for sharing your experiences! Could you perhaps give us a better idea how the map you describe looks like? Are you using formulas for calculations? How do you transform between the different models and especially to Freeplane?

Volker
boercher
 
Posts: 644
Joined: Tue Jul 26, 2011 7:13 am

Re: A Novel Use For Freeplane

Postby tylerreed » Thu Mar 27, 2014 8:26 pm

A quantitative mapping consists primarily of three types of nodes: FusionPoints, References and Metrics. The primary node type is called a FusionPoint and it acts as an aggregation point for its child nodes. Every FusionPoint has the attributes: Name, Calculation and Attribute (and sometimes others). Calculations currently supported include all basic logic functions (and, or, not, xor), Sum, Difference, Product, Quotient, Percent, Exists, Connects, Parent-Child and some strange, domain-specific things. The child nodes of a FusionPoint can be any of the other node types. In the case of References, these refer to architectural elements inside an enterprise architecture model. Say, for instance, a set of business processes.

The Attribute of a FusionPoint denotes which attributes of its children should provide values. For example, a FusionPoint with a SUM calculation and an Attribute of "TimeToComplete" would iterate through it's child nodes, accessing those elements in the architectural modal (stored in an EA app) and sum their respective TimeToComplete values. This would provide a total "Process Timeliness" value. In another example, a FusionPoint with a Calculation type of EXISTS would return true if the child node references all exist within the architecture.

Metrics always have two children, a FusionPoint and a Normalization table. The Normalization table is simply a node with a set of value-pairs as attributes. During processing, the analytic engine calculates the value of the Metrics' FusionPoint and applies it to the normalization set. This allows to put calculated values into a normalization scheme that is appropriate for the perspective we are using.

The entire map can be of any depth. FusionPoints can contain FusionPoints with Metrics that subsequently reference FusionPoints, etc, etc... ad infinitum. My working map currently has thousands of nodes with the deepest tree being maybe 16 nodes deep. I bought an additional monitor that I orient vertically so I can see more of the tree sections that are tall (lots of sibling nodes).

The entire calculation scheme is a depth-first recursion. The analytic engine (written in VB.Net) reads the map into its internal representation and then accesses the EA models via an API. It also has limited abilities to write out new Freeplane maps with results from the analytic process.

That's a general overview. Lots of features and side-bars not mentioned here.

Thanks for asking,
Tyler


boercher wrote:Hi tylerreed,

thank you for sharing your experiences! Could you perhaps give us a better idea how the map you describe looks like? Are you using formulas for calculations? How do you transform between the different models and especially to Freeplane?

Volker
tylerreed
 
Posts: 3
Joined: Tue Mar 26, 2013 2:30 pm

Re: A Novel Use For Freeplane

Postby kewapo » Fri Mar 28, 2014 8:40 am

I am a Knowledge Engineer and I am very interested in all kind of knowledge representation, from text, hierachical to concept maps, mind maps and ontologies.
I don't know anything about this quantitative maps.

Can you send us any in-depth information about them?


Thanks
kewapo
 
Posts: 14
Joined: Mon Apr 15, 2013 10:44 am

Re: A Novel Use For Freeplane

Postby tylerreed » Tue Apr 01, 2014 12:26 pm

Quantitative maps... they aren't a formal class of knowledge maps, it's just what we call them. We have 4 primary pieces: an ontology, an enterprise architecture model that adheres to the ontology, a quantitative mapping (in Freeplane) that describes aggregation schemes for different perspectives, and an analytic engine that processes maps against models. Each map represents a different quantitative perspective of the model, e.g. maturity, financial, governance, etc.

-Tyler
tylerreed
 
Posts: 3
Joined: Tue Mar 26, 2013 2:30 pm


Return to Open Discussion

Who is online

Users browsing this forum: No registered users and 1 guest