AKBC workshop ideas 2020

Title: OKNow what?

Topic: Infrastructure requirements for [supporting] OKNs

Questions to address, now that many OKN initiatives are underway:
+ What's known but missing, what needs more adoption, what's still an open challenge?  
+ What norms and practices are needed for different topic areas? 
+ When I contribute to an OKN, how can I know how widely it is accessible?
+ How to share namespaces for models, shapes, functions 

Outputs: Review+ catalog of current tools;  strengthening community across the networks.

<another 2 paras of description here>

<who should come + why>

Format: Full day.  

1 keynote, 2 invited provocations
6 invited papers w/5m talks, 6 invited working demos
A closing workgroup to compile a summary report.
An open call for papers and tools to include in reading materials. 

Potential program committee

Lara Campbell [NSF], Guha [Google], Mike, SJ, Lydia Pintscher...

Other potential participants

From our meta-libraries [Internet Archive] - Bryan Newbold?
From the decentralized web [IPFS] - 
From specialist graphs: 
   Social science + Individuals - Julia Lane's Rich Context grew (Dan Mbanga?)
   GIS + location data - ?
   Environmental data - ?
   Bio data [Broad Institute, Allen, +] - ?
   Publication + reference data [SemScholar, Lens, Meta] - Lucy Lu Wang
Knowledge graph + related tech:  
   Wikidata:  Magnus Manske [Wikidata+],  Kingsley Idehen [dbpedia]
   Extraction:  Sarah Chastens
   [Truth] evaluation: Duke Lab?
   BinderHub/Notebook ecosystem:  Yuvi Panda
   

Ideas for the session

Live demo stations before lunch:  full-track tutoriald and demos of existing work, lined up along workflow lines: from source to annotation to application programming.

Summary working group in the afternoon: compiling a report from the day w/ recommendations, via full-room discussion.  (or one breakout group per ~15 people)
+ What are the top five features needed to support OKN verticals? 
+ Walk through stages in the knowledge lifecycle:  sharing, contextualizing, sifting, mining, querying, disambiguating, aggregating

Adversarial panel: approaches to challenges of adversarial data environments, with disinformation, database and model poisoning, and more.

————

Old ideas

~ Sharing: Trace a new dataset through genesis, description, compression, publishing, indexing, notification, processing, updating

~ Sifting, coloring: Trace tags over time: identity, authority, review, accuracy, relevance, currency, other qualities

~ Mining: Trace refactoring and reuse over time: first-order extraction, second-order cross-correlation, third-order statistical extraction, updates

~ Finding: Trace a query through networks, time, space: as it is compiled, cached + mirrored, named + referenced, updated + versioned + annotated.

~ Prov: What does the profession of provenance enriching + confirmation look like, now and tomorrow?

~ Trust: What does an optimal high-trust network look like, for a constellation of well aligned participants?

~ Mistrust: What does a no-trust network look like, for a universe of personal knowledge clients protecting what you know or think you know, under adversarial conditions