Transcript forc-future-hovy-v3x

The Future
Or: What I’d Like from
Publications of the Future
Eduard Hovy
USC/ISI
www.isi.edu/~hovy
Vision
To improve the communication of knowledge
between scholars using new informatics
technology
Why publish?
• Record new knowledge (data, insights,
interpretations, theories…) with a time-stamp
• Communicate this to others
• Enable measurement of personal performance (e.g.,
for proposals or tenure)
• Train new scholars to write
A PoF should do all the
traditional things papers do,
but also more…
What PoFs add
• Better knowledge access
(help interested people locate the work)
Use terminology standards
Automate access
• Better knowledge communication
(help the reader understand the work)
Reflect the foundational
theory and methodology
Contextualize work in
relation to current world
Use best media at hand
• Better knowledge verification/extension
(penalize bullshitters)
Expose the reasoning
Provide non-text info and tools
So why don’t we do this today???
Why/when publish with a PoF?
(Why would our vision stick?)
• When the PoF is…easy to produce
– learn and use tools quickly
– capture all aspects of the knowledge
• Easy to consume
No impediments
– ingest the results faster and deeper
• Cheap
• More effective
– readers get more info, more accurately
– ditto for your proposal reviewers
– (tenure) evaluators get a more accurate picture
Many benefits
• Cool
– all other things being equal, making a PoF is cooler than not
What’s our problem?
• We’ve got an intuitive idea of what we mean
• We’ve got lots of cool tools and demos
BUT
• We don’t have a clear vision of what’s possible
• We don’t have a clear statement of why
• We don’t understand the impediments and costs
• We don’t have a procedure/environment yet
Goals of the meeting
• (Define the vision: To improve communication)
• List and organize the methods (what things can I do?):
– All kinds of functionalities to aid producers and consumers
– Example tools that illustrate them today
– Technology needs
• Study the impacts/effects (why is it beneficial?):
– On knowledge consumers
– On assessors (of proposals, of us)
• Study the financial implications (what will it cost
me?)
The report
• Improving knowledge communication
– Vision: What are the communication functionalities
and their integration
– Technology: What are the tools for doing this
• Impacting our world
– Social aspects: How do we quantify impact of use
– Coolness: How do we make it attractive to do/use
• Overcoming obstacles
– Financial considerations: How do we make it
sustainable
– Getting the ball rolling: How do we start?
VISION: THE BENEFITS
A PoF should do all the
traditional things papers do,
but also more…
Additional desiderata for a PoF
• Make the work easy to find
(automatically push toward the right
place; preemptively advertise)
Use terminology standards
Automate access
• Make it easy to ingest/understand
(help the reader, at all levels)
Reflect the foundational
theory and methodology
Contextualize work in
relation to current world
Use best media at hand
• Make it easy to verify/extend
(penalize bullshitters)
Expose the reasoning
Provide non-text info and tools
How to achieve this?
Contextualizing the work:
Ideas, papers, researchers, practitioners
• Goal: Situate the work in the evolving stream of
knowledge
• Contexts: The Web-of-X
–
–
–
–
Ideas / theories
Papers
Social network of researchers
Research network of practitioners (some fields have
theoreticians and practitioners: Economics–Banking,
Psychology–Therapy, PoliSci and Public Admin– eGovernment, etc.
Approach 1
Approach 2
Approach 3
Automated Citation Summarizer (Wan et al.)
Adding other media into text
Inserting more other-media into papers (O’Donoghue et al.)
Active pdf in documents (Greenacre and Hastie)
Non-text info and Tools
• Goal: Publish also the actual non-text data and tools
– Provide
• Data, images, movies, etc.
• Data collection and manipulation machinery and
software
linked to their actual mentions in the paper
• Maintenance
– Update, maintain, and extend this as new data becomes
available
• Who will do this?
– The argument: Publishers as Archives / Data Warehouses
…and let’s not forget presentations!
They are PoFs too, and should be
connected to the paper(s) and
contexts and termbanks,
and should also contain supporting
frameworks
The above is common to all (or most)
disciplines and fields
BUT the various types of data and the
various underlying theoretical
foundations and methodologies
differs…
PoFs should reflect/support
foundational theory and methodology
• Goal: Present the work in the form most ‘consumable’
by the expert reader
• Vision:
– Provide frameworks or templates reflecting the
underlying theoretical constructs and
methodologies of the field
– Encourage authors to conform to the template at
the crucial points
– Link terminology and procedures to standardized
and centralized termbanks
– Discourage free-form argumentation and BS
Ex 1: Measurement as foundation:
Scientific experiments
• Problem: How to describe causality in the world
• Example: KE-f-ED model for Biomedical experiments
(Burns):
– Independent (input) variables determine dependent
(output) variables
– Experiments are determinations of the (causal) relation
– Model:
•
•
•
•
•
Describe domain and hypothesis and variables
Describe experimental setup, incl. independent variables’ values
Describe measurement process
Describe output: dependent variables’ values
Discuss how this validates (or not) the hypothesis
KE-f-ED example (Burns et al.)
Ex 2: Balancing concerns as foundation:
Ethics
• Problem: How to reason about balancing concerns
• Examples: Actions of tobacco companies, chlorine
industry, nuclear power plants… — or governments
regulating them
• Foundational pattern: The Precautionary Principle
– In the absence of scientific consensus, the burden to prove
harmlessness lies with the agency wanting to act
– Algorithmic formalizations to support discussion and
quantification of Risks (probabilities of Bad Things) and
Hazards (intensities of their effects)
– Research papers: Case studies giving situations, economic
and social and environmental effects, numerical data, etc.
PoFs on Precautionary Principle
• A PoF should include
a framework or
template of the
reasoning scheme,
into which the author
must/can cast the
argument
Ex 3: Evolution over time as foundation:
Musicology and History
• Problem: How to understand temporal development of
things: music, political states, societies…
• Basic foundational pattern: Characterization of principal
features at various points in time, with (causal) linkages
between them
• Example: Evolution of understanding of harmony:
– Formalization of increasingly sophisticated models of harmony
from pre-Bach (’horizontal’ melodies in modes) to Bach (‘vertical’
chords of notes)
– Foundation: Timeline, with data and reasoning in multiple
modalities linked into it
• Music scores, actual music, past theoretical works…
IMPACTS OF POFS: SOCIETY
COSTS OF POFS: FINANCIAL ASPECTS
What are we paying for?
Web
Reviewing
Recruiting
Indexing
Typesetting Reviewing
Storing
Copyediting Typesetting Delivering
Copyediting
Advertising
Distributing
Accessing
Blogs
Living in the future
• If you make available a wonderful PoF-building and
distribution system, with open access, what then?
• The immediate future would be great
• And the longer-term future?
–
–
–
–
Code no longer works
Data grows in incompatible ways
Fancy media are not supported any more
Formats go out of date (troff, Scribe…)
The Business Case
• Somebody has to do the work we don’t want to:
–
–
–
–
Authoring tool standardization and support
PoF copyediting and typesetting and formatting
Data archiving and updating
Data software archiving and support
• Somebody has to pay for this:
– The producers can pay some
– The consumers can pay some
• It would be nice if we knew the individual costs and
how bundled they necessarily are
Where should we go from here?
• Double-publish: Paper-lite and Paper-rich
– Measure effort and cost to produce both versions
– Measure consumption of both
– Find where’s the value… What do people ‘pay’ for?
• Get the publishers involved: show them where they
can play a role
– (And perhaps funders? But better if not…)
• Get more people (and humanities researchers)
involved