On the horizon:
What is Solid today, and where could we be headed?

About Me

• Built some Semantic Web applications @ ANU
• Built some RDF Reasoners for the Web @ ANU
• Built Solid Apps @ ANU SII
• Working on Solid @ Inrupt and with SolidLab
• Will also work on Solid with EWADA Team @ Oxford

⚠️ Disclaimer

• Opinions expressed here are my own.
  
  • They do not necessarily reflect the current roadmap of any organisations I am affiliated with.
• I am not the editor of the spec.
  
  • This presentation serves as a position statement, and is not an official statement of the Solid spec.

Overview

• Solid Today
• Solid views: Towards Solid as a contextual KG
• Back(door) to the future: Extension APIs for Solid
• Solidifying Solid: Standardising extension APIs
• Envisioning the far future: A Web of Rules and Data

A Solid foundation: What do we have today?

• LDP "a file system for the Web"
  • Uses HTTP(S)
  • Performs CRUD operations on resources (documents)
  • Resources can be RDF & non-RDF
  • Resources are organised in containers (folders)
• Authorisation (ACP / WAC)
  
  • Permission agents & applications to read, write, append and delete resources

What are people doing with Solid today?

• File Browsers
  • SolidFS
  • PodPro
  • PodBrowser
  • Penny

What are people doing with Solid today?

• Web Apps
  • BBC Together (blog post)
  • itsme (announcement)
  • Media Kraken
  • Umai
  • Solid Focus
  • 
  • 🍜 KNoodle

What are people doing with Solid today?

• Query
  • Comunica with features including:
    • Link Traversal https://trends.knack.be/ondernemen/itsme-en-het-vlaams-datanutsbedrijf-athumi-willen-met-datakluizen-europa-veroveren/
    • Reasoning
    • GraphQL-LD
  • GraphQL Experiments

Where are some difficulties?

• Interoperability
  • Applications cannot understand data written by other apps
  • Applications cannot find data written by other apps
• Performance
  
  • Apps struggle to work with "big data"
• Requesting & automating authorisation

What solutions can we use to find data today?

• client-to-client agreements
  • Type Indexes
  • Closed ecosystem assumptions
• brute-force discovery
  • Pod-wide link-traversal
    • Use tools like Comunica
    • Optimize with services like the ESS Fragments Indexer (but don't rely on them)

What solutions can we use to understand data today?

• client-to-client standards
  • agree on shapes
• advanced mode
  • client-side reasoning with:
    • eye
    • N3.js
    • Comunica

What are shapes?

PREFIX schema: <http://schema.org/>
shape <#PostalAddress> {
    schema:streetAddress schema:Text [1..1]
    schema:addressLocality schema:Text [1..1]
    schema:addressRegion schema:Text [1..1]
    schema:addressCountry schema:Country [1..1]
    schema:postalCode schema:Text [1..1] pattern="^\\d{5}(-\\d{4})?$"
}

How can you help today?

• Applications
  • Build them!
• Shapes
  • Build shapes
  • Share on a shapes repository
  • Reuse from an editable or read only shape repository

What is a shapes repository?

What is a shapes repository?

What is a shapes repository?

Which difficulties can we solve today?

• Interoperability
  • For finding data we have type-indexes, but they can be corrupted as they are client-maintained
  • For understanding data we can agree to use the same shapes; but we do not access the power of RDF semantics
• Performance
  
• Requesting & automating authorisation

What's on the horizon?

• Interoperability
  • Data views such as
    • a server managed type index
    • virtualising data by shape
• Performance
  
  • Data views for query optimization
• Requesting & automating authorisation
  • Auxilliary resource views on "interesting" policies

What are these views of?

• Short term
  • The LDP storage
  

What are these views of?

• Long term
  • A Hybrid-Contextualised Knowledge Graph containing:
    • Application Data
    • Provenance (metadata)
    • Policies
    • Rules
  

• Solid Today
• Solid views: Towards Solid as a contextual KG
• Back(door) to the future: Extension APIs for Solid
• Solidifying Solid: Standardising extension APIs
• Envisioning the far future: A Web of Rules and Data

A seamless transition

A seamless transition

• Data integration not API integration
  • Reflections of knowledge
• Your LDP API will still work!

A seamless transition to data integration

Back(door) to the future: Extension APIs for Solid

• Policy APIs such as:
  • Retention periods
  • Access Grants
• Reasoning APIs:
  • Shape/vocab alignment
  • Domain inference
  • Custom rules

Which difficulties can we solve today?

• Interoperability
  • For finding data we have type-indexes, but they can be corrupted as they are client-maintained
  • For understanding data we can agree to use the same shapes; but we do not access the power of RDF semantics
• Performance
  
• Requesting & automating authorisation

• Solid Today
• Solid views: Towards Solid as a contextual KG
• Back(door) to the future: Extension APIs for Solid
• Solidifying Solid: Standardising extension APIs
• Envisioning the far future: A Web of Rules and Data

Solidifying Solid: Standardising extension APIs

Danger ahead!

• Take the following slides with a grain of salt
• Unlike the rest of the presentation, these ideas live only within the academic community
• We are now talking potential developments the far future (at least 10+ years).
• I have a personal fondness towards RDF reasoning engines.

Envisioning the far future: A Web of Rules and Data

• No application logic
• No code
• Just a (socratic) reasoning dialogue

What does this dialogue look like?

• Set a goal
  • I want my fridge to be well stocked, with healthy food that I like, whenever I am home.
• Communicate (using RDF over HTTP) your goal, or sub-goals to relevant stakeholders.
  • By sending questions formulated in RDF from your reasoning engine to reasoners hosted by the Supermarket and Delivery Service.

What does this dialogue look like?

• Stakeholders respond with partial answers, or questions, permitted by their policies.
  • Here are the prices of the items you are interested in. Please tell me the time you want to pick it up so I can tell you the stock I will have at that time.
• Rince & repeat

Envisioning the far future: A Web of Rules and Data

• • "A few decades ago, software took over many jobs from hardware: calculations that used to be hard- wired became soft-wired as lines of code.
    
    [...] Similarly, reasoners will start rewiring software for specific situations. EYE already does this when performing rule-based Web service composition."
    
    - Jos De Roo

Back on track

If you have one takeaway...

• Build apps and shapes today!