Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Memory Connected Extending Memory on the Web via Human-Centric Knowledge Exchange Network
Jie Bao and Li Ding
1
Memect Memory Connected
Memect Memory Connected
Big data is right here, created and used by human
2
http://www.gooddata.com/images/uploads/big-data-image.jpg
Memect Memory Connected
Social Web and Walled Gardens
• I’m locked in a garden • My brain is fragmented – (Dion’s talk) • Loyalty V.S. Trapped
• I don’t use a garden • Prefer not: bank statement • No such service
3
person A Person B Silo Y
Silo X
Memect Memory Connected
Human Centric Design – Principle 1
4
X
Human readable knowledge • To extend human’s memory • To facilitate human communication
Memect Memory Connected
Human readable knowledge: Design Issues
• For developers • Cognitive complexity over computational complexity
• “knowledge must be written for people to read, and only incidentally for machines to understand” *
• Readable code and data modeling (cf . Yaml, python, coffee-script, groovy, turtle etc)
• Engineering friendly: reusable, module support • No need for 100% accuracy, some guess is fine
• For users • Readable representations • “right” granularity • Explorable knowledge with just-in-time schema
5 * Abelson and Sussman claimed that “programs must be written for people to read, and only incidentally for machines to execute”
Memect Memory Connected
Human Centric Design – Principle 2
6
Tangible rewards and cost reduction • Instant rewards • Cheap start
Book Blog Tweet
Memect Memory Connected
Tangible rewards and cost reduction: Design Issues
• For developers • Reduce upfront agreements (ontology, schema etc) as
much as possible, e.g. avoid extended scrum.
• Allow mixed representations, allow data/schema to evolve and improve
• Ownership, not blocked by mandatory consensus
• Keep it simple
• For users • just-in-time validation and refinement, e.g. speadsheet
7
Memect Memory Connected
Human Centric Design – Principle 3
8 PROVO ontology draft (2011), http://www.w3.org/TR/2011/WD-prov-o-20111213/
Bottom-up growth scales • Top-down consensus may take years
• By experts • Bottom-up: shared and evolving
• By users
Memect Memory Connected
Bottom-up growth: Design Issues
• For developers • Instead of invent new, build on top of well-adopted
technologies, e.g. json, csv,
• Live schema from live data
• For users • More on Web data
• More on everyday-life, consumer data
9
Memect Memory Connected
Use-cases
• Support mailing list users • Daily routine: don’t miss signal, don’t waste time
• Conversation: need timely reply
• Research(task): life cycle of working group action item
• Research(user): build trust in to unfamiliar users
• Support team common knowledge • Consensus: Is it just the project manager’s view?
• Unexpected use: what if one member left? How much should I pay for unforeseeable future?
10
Memect Memory Connected
Vision: Knowledge Exchange Network
• Extended Memory: capture and organize
• Connected Memory: exchange
11
person A
Person B
person C
Broker X
Broker Z
Broker Y
Questions? Comments?