This work builds on the "Graph-Based Data Science" talks at conferences and meetups, which experiment with the content – collecting feedback, critiques, suggestions, etc.
Feedback from those interactions and experiments get integrated into these materials and eventually land here. To illustrate:
The objective for these learning materials is to help people learn how to leverage kglab effectively, gain confidence working with graph-based data science, plus have examples to repurpose for their own use cases.
You'll find a mix of topics throughout: data science, business context, AI applications, data management, data strategy, design patterns, distributed systems – plus explorations of how to leverage the math, where appropriate.
The intention is to make these materials useful to a wide audience. So we provide multiple entry points, depending on what you need...
A Grammar of Learning¶
Let's talk briefly about how to design and present learning materials. What are we doing here? In another sense, consider the following as a very brief exercise in design thinking, focused on how to improve software documentation and tutorials.
As described in the background below, there's a growing movement toward more structured approaches to how documentation gets presented. Arguably, a lack of quality or effectivenes in documentation creates major impediments for open source libraries. Much the same goes for tutorials, which are often written from the perspective of someone who's already familiar with the material – thus defeating its purpose for those who aren't.
Our team has iterated on these approaches, blending key insights and practices from prior roles, to create design patterns for documentation and tutorials. The outline of this documentation follows these patterns.
Design affordances for navigation, indexing, search, and recommendations.
Much effort has gone into making sure this material is thoroughly hyper-linked together, to help with both navigation and indexing. Even so, suggestions for improvement are highly welcomed!
The left sidebar provides an overall outline of sections, while the right sidebar links to sections within the current section.
Start at any point, for whatever info is most immediately useful.
It may be helpful to run JupyterLab for the coding examples in one browser tab, while navigating through this documentation in another browser tab.
The search features implemented in documentation frameworks such as
etc., leave much to be desired.
Consequently the search features have been disabled here, for now,
while a replacement is in development.
Meanwhile, contextual links based on the KG-powered Derwen
search/discovery services help provide
More of that is getting integrated directly into this documentation.
Present problem-oriented directions for getting things done.
- Getting Started with enough code to begin using the library in use cases.
- Specific "HowTo" articles, generated from notebooks.
See also https://diataxis.fr/how-to-guides/
Learning-oriented practice through hands-on coding exercises.
- Tutorial generated from notebooks, which provide reusable sample code plus supplemental exercises.
- using a progressive example.
- including a learning promise styled syllabus.
See also https://diataxis.fr/tutorials/
Provide explanations that introduce concepts, exploring the theory and processes behind their usage.
- What's a Knowledge Graph?, attempting to get shared definitions in place – for a subject that has been somewhat difficult to define.
- Graph Concepts – exploring the abstractions used in this library, leading to best practices for how to leverage it.
See also https://diataxis.fr/explanation/
Where does this kind of technology meet business needs?
- Use Cases exploring case studies for KG use in industry applications.
- Other industry analysis, articles.
Each type, class, and parameter in the library which is intended for public use gets listed and explained, with plenty of whitespace for readability.
- Package Reference generated from apidocs customized for improved readability, corrected type annotations, etc.
- Dependencies links to all required libraries, with caveats.
- Build Instructions for building from source (not necessary in most cases).
See also https://diataxis.fr/reference/
Directing people toward how to engage with the developer community and related support.
- Community Resources for more information about using this library, troubleshooting issues, and getting involved as a contributor.
- Acknowledgements to the developers, sponsors, and organizations which help to support this project.
Clarify terminology with shared definitions and their supporting links, and identify original papers, histories, notable authors, and other materials for context
- Glossary, which define terms and also provides a basis for indexing.
- Bibliography of primary sources cited within this work.
Helpful feedback, both from the community to the developers, and also for learners who are new to working in this field.
- surveys for feedback about improving the learning material
- self-assessments for personal feedback (WIP)
- certification exam (WIP)
- coding examples that lead into a capstone project (WIP)
- instructor's personal recommendation on social media, following successful completion of the above
The strategy for structuring documentation which is used here has been guided by some highly recommended resources. Kudos to @louisguitton for pointing out these practices and initiating this dialogue for kglab.
Diátaxis, Cloudflare, Stripe, and other organizations have fostered writing cultures internally, while applying more structured approaches to presenting their learning materials externally. That shouldn't be surprising: software developers tend to write in volume – generally much more text than they produce as source code – since the processing of planning projects and maintaining them over time usually involves lots of written communications. That said, good editing for developers' text is much less common. That's why the examples set by Stripe and others are important.
First, check out [victorino2020] for a description of Stripe:
Their strong writing culture benefits the organization in several areas:
- Time efficiency. Sharing ideas through writing eliminates the need for repetitive verbal updates to disseminate ideas and information.
- Knowledge sharing. Documenting important ideas forces clarity of thought and makes information more accessible to everyone in the company, versus slide decks that are ephemeral and require less rigor of thought.
- Communication. Clear writing requires clear thinking, meaning employees invest more time shaping their ideas before sharing them.
The majority of the guidance comes from [procida2020] at Diátaxis:
There is a secret that needs to be understood in order to write good software documentation: there isn’t one thing called documentation, there are four. They are: tutorials, how-to guides, technical reference and explanation. They represent four different purposes or functions, and require four different approaches to their creation. Understanding the implications of this will help improve most documentation – often immensely.
In particular, the implementation of this approach by [song2020] for Cloudflare documentation served as a key inspiration.
The notion of a grammar of graphics was introduced by
then implemented in the
To some extent, this work can borrow from that, attempting to
articulate design patterns to be used as a grammar of learning.