Aggregating Operational Knowledge in Community Settings
1. Aggregating Operational Knowledge
in Community Settings
Srinath Srinivasa
Open Systems Laboratory
IIIT Bangalore
India
sri@iiitb.ac.in
http://osl.iiitb.ac.in/
5. Commercial Clusters
● No organizational structure
● Individual shop owners join cluster autonomously
● No overarching reporting structure
● Collective action taken by consensus
● More organized than a crowd
● All shop owners have something in common
● Shared interests to collaborate and compete
● Communities: Generalization of Commercial Clusters
12. Encyclopedic Knowledge
Aggregates several local
perspectives to a global whole
Encyclopedic knowledge
A convergent process of
aggregation
No subjective versions
Local perspectives
Quality based on balancing
POVs
15. Operational Knowledge
Utility 3 Aggregates a set of common
knowledge into different local
utilitarian “worlds”
Well known
Common Subjective by definition. User is
Knowledge a part of the encoded
knowledge rather than an
outside observer
A divergent process of
Utility 1 Utility 2 “aggregation”
16. Operational Knowledge
● Most common to dynamics of communities
● Concerned with putting a set of common knowledge to
different uses
● Subjective by definition: what is utilitarian to one need not
be utilitarian to another
● User (consumer of knowledge) part of the encoded
knowledge base rather than an outside observer
● A divergent process: communities necessarily dilute their
common condition by utilizing it in different (interrelated)
ways
17. Aggregating Operational Knowledge
Essential requirements of operational
knowledge app:
Support a divergent phenomena with minimal
redundancies
Support mechanisms to fill cognitive “holes” in a
divergent process
18. Many Worlds on a Frame (MWF)
● Proposed data model for capturing a divergent
knowledge aggregation phenomena
● Partially implemented in an application called RootSet
(http://rootset.iiitb.ac.in/)
● Expressible as a superposition of two modal Frames in
Kripke semantics (a posteriori analysis)
20. MWF: Frame
Concept hierarchy Containment hierarchy
Inherits properties, Inherits privileges and
associations and world visibility
structure Rooted in a concept called
Rooted in a concept called UoD
Concept
21. MWF: World
A world is a concept that can
is-in is-a University host relationships between
concepts and host “Resources”
(Files, Media, Web links, RSS
feeds, etc.)
Department Course
Org Unit Activity
Concepts participating in a
world are “imported” from the
Faculty Frame and play a “Role” in the
Student World
Person
Person
Roles are connected with one
another with “Associations”
22. MWF: Instances
● Any concept that cannot be subclassed is
called an Instance
● In any instance of a world, a relationship
instance can be added between two concept
instances, iff a relationship type exists between
the respective concepts in the world type
ancestry
23. MWF: Privileges
● Users and privileges an integral part of operational knowledge
● MWF privileges broadly ordered into following levels:
● Frame-level privileges
● Structure-level privileges
● Data-level privileges
● Visibility privileges
● Privileges are inherited through the is-in hierarchy
● A user having privilege p in concept C will have a privilege at
least p in all concepts contained in C
24. MWF: World Creation
New worlds can be created in the following
ways:
● Simpliciter
Create and manually specify lineage (is-a, is-in ancestry)
● Clone
Create new world with same structure and is-a ancestry,
specify is-in ancestry manually
● Induce
Create new world within an existing world by inducing a
new world around a part of the structure. Specify is-a
ancestry manually
25. Cognitive Gaps
● Divergent phenomena entails knowledge base forking
off in different directions
● Diversified attention
● Reason for communities to be less efficient than
organizations
● Possibility of emergence of “Cognitive gaps” --
elements of knowledge that get left out because
attention is diversified
● Need for Cognitive “gap fillers” -- semantic
recommendations by the knowledge base
26. Cognitive Gap Fillers
Heuristics to suggest knowledge elements to fill
cognitive gaps:
Data level heuristics
● Principle of locality of relevance
– Instances that play a role in a world are typically found in
the vicinity of the world itself
● Birds of a Feather principle
– Similar instances play similar roles in similar worlds
27. Cognitive Gap Fillers
● Data level heuristics
● Resource diffusion principles
– Resources in a world are typically relevant to concepts that
play a role in the world
– Resources held by a concept playing a role in a world are
typically relevant to other concepts playing similar roles
● Structure level heuristics
● Triadic closure
– If concept A is related to concepts B and C in a world, the
greater the strength of the association by virtue of number of
instances, the greater the possibility that B and C are
semantically related
28. Cognitive Gap Fillers
● Structure level heuristics
● Clustering principle
– Concepts tend to form semantic clusters where
association among elements of a cluster are tighter than
associations across clusters