Agile Practices for Knowledge Graph Engineering

Event Details

LocationRoom 1 (Auditorium)
DateMay 3, 2022
Time10:30 AM to 12:00 PM

The goal of this workshop is to gather ontology engineering and semantic web knowledge graph practitioners for an in depth discussion on the topic of agility in building EKGs (Enterprise Knowledge Graphs). The agile movement transformed the software development process from a linear, waterfall model to an efficient feedback loop by taking existing coarse grained practices and artifacts, turning them into smaller scale versions of themselves and defining organizational patterns that can execute short delivery cycles. 

How can we do the same for knowledge graph development?

Let’s share lessons learned,  success stories, practices, ideas, pitfalls and general accumulated wisdom. 

After a 10min introduction of the topic and the structure, attendees will form discussion groups of sizes 5-10 for 65 min discussions, before concluding with key takeaways from each group in the last 15min. 

Each group discussion will be structured as follows:

  • Introductions and informal discussion – 10-15 min
  • Topic discussion – 40 min
  • Summary – 10 min

Each discussion group will address the classical phases of an agile iteration cycle: 

requirements -> planning -> design -> development -> testing -> deployment -> review

And attempt to relate them to concrete activities, skills and artifacts that come into play during knowledge graph engineering. Suggested topics and questions to address:

  • Story definitions and how they differ according to the type of KG (data integration, ML etc.
    • What are best practices to define good stories? 
    • What’s the definition of done?
  • How to best “divide and conquer” ontology engineering?
  • What does it mean to embrace the core values of the Agile Manifesto in KG development?
  • What is similar and what is different b/w knowledge graph engineering vs. software engineering? 
  • What determines the cost of change and how can it be mitigated?
  • What does the ideal knowledge graph team composition look like? 
    • What does it mean to be a product owner for a knowledge graph? 
    • How to engage with business stakeholders in the absence of GUI mockups? 
    • Should subject matter experts be embedded in the team?
    • Knowledge elicitation – interaction between KG experts vs SMEs.
    • Competency question (CQ) elicitation – interactions between knowledge engineers and business/product owners.
  • Does the “do the simplest thing possible” philosophy apply here?
  • Iteration, feedback and validation – what is a good increment/deliverable? 
  • How to validate a deliverable and what to demo on demo day?
  • Tooling
    • What are good productive tools that help with agility?
    • What’s missing from the current toolset in the market at large?
    • How do you do refactoring for knowledge graphs?
    • How do you deal with cross-cutting changes – from ontology, to data mappings and queries in downstream code?
  • SemOps: what does CI/CD look like for knowledge graphs?

We strongly encourage prospective participants to contact us with suggestions of open questions, topics or tweaks to the proposed format.

All materials produced (notes, postit, summaries, open questions) will be collected and the organizers will assemble them in digital form to share with the community at large. Feel free to take pictures during the workshop.

Contact Us