I would like to test a set of interconnected ideas.
1. Transforming Theory of Change into a Knowledge Graph
A project's Theory of Change (ToC) can be transformed into a Knowledge Graph. A Theory of Change has a number of key entities: Activities, Results, Assumptions. Each can be characterized in some detail with indicators, properties, etc.. and there are clear relationships. An activity "contributes to" a result. An assumption can "influence" an activity. The ToC is a hypothesis based on past evidence from similar or related projects.
The challenge here may be to ensure that the ToC's complexity doesn't lead to an overly complicated Knowledge Graph. It will also be important to consider how to represent uncertainties and evolving assumptions over time within the Knowledge Graph.
2. Accumulating Data to Confirm or Adjust the ToC
As the project is implemented, data accumulates that will either confirm the ToC or require some adjustments. That's what adaptive management should be about. The data can combine structured, quantitative data based on predetermined indicators, AND unstructured data from learning activities such as After-Action-Reviews, interviews, stories, etc. All this data can be added to the Knowledge Graph.
The challenge here will be in consistently capturing and integrating unstructured data in a way that it's meaningfully represented within the Knowledge Graph. NLP tools may be necessary. Would they be effective? In addition, as the graph grows in complexity with new data, will it remain manageable?
3. Integrating KGs across Projects for Organizational Learning
Across an organization, tools like Propel that create a structure for knowledge capture could lend themselves well to integration with a Knowledge Graph to aggregate data across projects AND enable more sophisticated analysis. A set of project-based Knowledge Graphs developed based on their individual ToCs (as in steps 1 and 2 above) can be connected into a bigger Knowledge Graph. At first, this expansion could be focused on a particular sector with a "Grand Sector Theory of Change" with regional as well as country-specific adaptations and tailored assumptions based on context.
Challenges might include a) interoperability between different project Knowledge Graphs, particularly if they were developed independently or using different methodologies; b) the requirement for a unified ontology or framework for creating these graphs; c) associated governance and data-sharing protocols to enable integration without compromising data integrity or security.
4. Portfolio-level Analysis
Expanding the scope further, a donor could look at an entire portfolio of projects within a comprehensive knowledge graph, combine insights from the graph with meta-analysis like evidence-mapping reviews to help generate better RFPs. That would still require extensive country/local knowledge.
Challenges might include scalability issues, particularly in ensuring that the Knowledge Graph remains up-to-date and relevant across a broad portfolio. The need for documenting extensive local knowledge is also a potential bottleneck.
5. Local Knowledge Mapping and Knowledge Graphs
Therefore, local knowledge mapping and knowledge graphs would need to be further developed for and by local SMEs to provide leverage for local organizations to impact project designs and project implementations.
The challenge might be in creating incentives and building capacity for local SMEs to develop and maintain these Knowledge Graphs. There would be issues related to standardization to ensure that local KGs can interface with broader organizational or donor-level KGs without losing the richness of local insights. There would also be significant issues related to data ownership.
This is potentially unrealistic, overly ambitious, crazy, etc.. I wonder if anyone has already started working on step 1 because I often think I've come up with an idea and soon realize it's already been done (or tried). Turning a Theory of Change into a Knowledge Graph is definitely feasible. Then we would discover whether that's useful at all, before thinking too much about the more ambitious steps.
No comments:
Post a Comment