Social Service Selection ServiceOriented Computing Semantics Processes Agents Munindar P Singh and Michael N Huhns Wiley 2005 Chapter 20 2 ServiceOriented Computing Semantics Processes Agents ID: 200854
Download Presentation The PPT/PDF document "Chapter 20:" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Chapter 20:Social Service Selection
Service-Oriented Computing:
Semantics, Processes, Agents
– Munindar P. Singh and Michael N. Huhns, Wiley, 2005Slide2
Chapter 202
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Highlights of this Chapter
Reputation Mechanisms
Recommender Techniques
Referrals
Social Mechanism for Trust
IdentitySlide3
Chapter 203
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Recommending Products vs. Services
Products (by a product vendor): often,
The recommender is the provider
Votes are known to the recommender
Votes are received prior to usage (buying)
Repetition is less likely (buy the same book?)
Services (by a service registry)
The recommender is not the provider
Votes are not necessarily known to recommender
Votes are given after usage
Repetition may occur, invisibly to registrySlide4
Chapter 204
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Reputation
Computationally treated as centralized
The agency (e.g., eBay) is the
central authority
that
Authenticates users
Records, aggregates, and reveals ratings
Provides the
central conceptual
schema for
How to capture ratings (e.g., numbers and text)
How to aggregate them
How to decay them over timeSlide5
Chapter 205
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Service Communities
Each principal
Provides services to
each others
Provides recommendations to others
Exploits services provided by others
Has a reputation
The agents assist their users and other agents in
Evaluating the services and referrals provided by others
Maintaining models of acquaintances
Deciding whom to contact for a serviceSlide6
Chapter 206
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Social Networks and Referral Chains
Referral chains in networks provide:
Way to identify a good provider
Way to judge the quality of a provider
Reason for a member to respond in a trustworthy manner
As the chains get longer
The trustworthiness of a recommendation decreases
The effort to find providers increases
Therefore, shorter chains are
better
Technical challenge: how can we find such?Slide7
Chapter 207
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Distributed Treatment of Referrals
Receive request
Model asker
Respond
Ask
Follow referrals
Use
Rate; updateSlide8
Chapter 208
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Agent Model for Referrals
Model each agent via its
Interest (services sought)
Expertise (services provided)
Models of its acquaintances representing their
Expertise (ability to provide good services)
Sociability (ability to provide accurate referrals)Slide9
Chapter 209
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Reputation Buildup and Collapse
A participant who begins to misbehave is detectedSlide10
Chapter 2010
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Small World Phenomenon
Milgram (1967): two individuals chosen at random in the U.S.A. are linked by a chain of 6 or fewer first-name acquaintances (empirical observation)
Six degrees of separation
Erdös numbers
Diameter of the connected Web: 20Slide11
Chapter 2011
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Small-World Network
Generated by perturbing a regular ring
A highly structured (clustered) network with just a few random edges
Random edges correspond to shortcuts
Yields high clustering and short paths
Direct relationships between agents who primarily belong to distinct subcommunities
Shortcuts: weak tiesSlide12
Chapter 2012
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Quality Relates Inversely to ClusteringSlide13
Chapter 2013
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Weak Ties versus Clustering
Conventional approaches recommend based on preferences of similar users
But, it is better to ask dissimilar people who bring a novel perspective
Define a form of
controlled scatteringSlide14
Chapter 2014
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Link Analysis
Web link: recommendation by page author
An external criterion for estimating the value of a page to others
Typically, web engines crawl the web, build giant indexes, and analyze links
Referral: dynamic, targeted recommendation by an agent
Similar mathematical concepts to above applySlide15
Chapter 2015
Service-Oriented Computing: Semantics, Processes, Agents
- Munindar Singh and Michael Huhns
Chapter 20 Summary
Selection should be
empirical:
based on data
Centralized reputation mechanisms gather data but impose many restrictions
Research Challenges
Social
networks can avoid
the above limitations
Referrals help maintain distributed social networks
Social structure can evolve collaboratively
Services can be rated and selected and rated again, and so on …