Flexible Execution of E-commerce Protocols: A Commitment ...

Flexible Execution of E-commerce Protocols: A Commitment ...

Referral Systems: Applications in Knowledge Management; Emergent Properties Munindar P. Singh (joint work with Bin Yu, pInar Yolum (mainly), Yathi Udupi; Yonghong Wang) Department of Computer Science North Carolina State University Outline Motivation and Framework Making Referral Systems Useful Understanding Referral Systems Authoritativeness Structure Evolution

Directions Backup Clustering Power-law networks 2 Knowledge Management Management of knowledge at the individual and enterprise levels The traditional approach is artifact centric: focuses on documents mainly Major shortcoming: most valuable knowledge is not in artifacts Ownership: opposed to individual interests Lack of context: where applied Violation of privacy: how much would you

reveal Need know-how: not just know-that Instead find people who know 3 Abstraction: Referrals for Selection How can we find a business partner in a purely distributed system? An agent represents a principal A offering or searching for services C An agent generates a query for a Q A service; sends it to its neighbors D

B Each neighbor may provide the R(D) Q service or refer to other agents (based on its referral policies) Each agent models the expertise (quality of a service) and sociability (quality of the referrals) of its acquaintances Based on these models, each agent can change its set of neighbors (using its neighbor selection policy): locally, autonomously Social network: as induced by the neighborhood relation Q 4 Why a Decentralized Approach? Problems with central authorities (e.g.,

Verisign) or reputation systems (e.g., eBay) Context and understanding: The contexts of usage may differ Empirical basis: Best to trust experience Did Verisign itself buy DVDs from Amazon? Privacy: Raters may not want to reveal true ratings in public Trust: Users of ratings dont necessarily know where the ratings come from 5 Motivation Referrals for service selection Follow referrals from trusted parties Self-organize based on previous

interactions Web structure Properties of its snapshot Stochastic models for approximating in-degree distributions Hyperlinks are assumed to be endorsements Local interactions are not captured 6 Referral process is explicit; emergent structure is not known to anyone

Emergent structure is known at the index; underlying process is not modeled Application Domains Commerce: C 1 Distinct service producers and consumers Producers have expertise, consumers have sociability Answers are easy to evaluate

Expertise of consumers does not increase C 2 C 1 C 2 C 3 C4 C7 C8 C9 C 5 7 C6

C 10 C 3 C 4 S 1 S 2 S 3 C 5 C 6 S 4 Knowledge Management: All agents can be producers and consumers

Answers are harder to evaluate Expertise of consumers may increase (expertise of the producers can be cached by others) MARS: MultiAgent Referral System Prototype system for helping people participate in a referral network Practical challenges: UI: use an IM client Communication: use an IM server (Jabber) Bootstrap: Infer peoples expertise and (initial) neighbors: mine email Research challenges How to evaluate convincingly?

8 Developed over several years by Bin Yu Wentao Mo Paul Palathingal Subhayu Chatterjee Ongoing Arvind Viswanathan Naren Ranjit (overlay on Facebook is planned)

Representations: 1 For simplicity, use vector representations for queries and knowledge Assume dimensions; supply values [spicy, timely, tasty, authentic, healthy]: [0.8,0.7,0.9,0.8,0.1] Vector Space Model Originated in the 1960s Still used in text retrieval Easy approach conceptually

Common in text retrieval Supports caching results But has well-known limitations 9 Applied by Yu & Singh; Yolum & Singh; Udupi, Yolum, & Singh Representations: 2 The meanings of the dimensions are not standard Ontology (loosely, conceptual model) for qualities of service

Common QoS: price, availability Domain-specific QoS: spiciness Idiosyncratic QoS: enjoyment How to handle preferences Decision theory 10 Maximilien & Singh; Maximilien developed a practical framework for QoS in Web services QoS framework works as a

reputation system; not yet combined into a referral system Propagation of Trust Referrals support trust management Provide a basis for finding witnesses, who can offer evidence (pro or con) about a third party Provide a basis for rating such witnesses Support adapting to select the more promising witnesses and avoid those who are deceptive

11 Yu & Singh: Applies DempsterShafer theory of evidence and weighted majority learning Wang & Singh: Theory of trust discounting and aggregation Analysis Not just develop a system and hope it works, but understand

its functioning to: Improve its effectiveness in important settings Find new uses for it Study general questions of the consequences of decentralization and emergence 12 The completed work has mostly had an empirical flavor Theoretical aspects would be great topics for further research

Referral Policies Refer all neighbors: Does not consider which neighbors would be more likely to answer (similar to Gnutella) Refer all matching neighbors: Refer those neighbors with sufficient expertise Refer best neighbor: Refer the most capable

neighbor. Guarantees that at least one neighbor is referred 13 Efficiency of Referral Policies Policies: Refer All Refer All Matching Refer Best Efficiency = # of good answers # of contacted agents Too many agents are contacted

Not enough good answers are found 14 Effectiveness of Referral Policies Low quality even though answers are found 15 Low efficiency but high quality Authorities Link analysis to find authorities from Web crawls

PageRank: Pages pointed to by authorities are also authoritative Factors that influence the emergence of authorities 16 P(i): PageRank of i N(j): Neighbors of j K(i): Pages that point to page i d: Damping factor Referrals and Authorities Web search engines

Mostly crawl static pages Interpret each URL as an endorsement Mine centrally to decide where to direct searches by all users Referral systems A decentralized agent Obtains dynamic (custom) information Knows if it is an endorsement Decides how to use it for its user Reveals appropriate information to others

Mining is optional, after the fact, for study and tuning 17 In referral systems, mining is used as a research tool Cannot centrally crawl a referral system in practice Exposing mined results may violate privacy Yolum & Singh Emergence of Authorities through Adaptation

Authorities emerge as agents change neighbors 18 Authoritativeness & Number of Experts 19 When the population has fewer experts, the authoritativeness of the experts is higher Effect of Referral Policies When more referrals are exchanged, the authorities obtain higher PageRank (i.e., their authoritativeness is greater) 20

Neighbor Selection Policies How do the agents choose their neighbors? Providers: Sociables: Weighted Average: 21 Choose the best m agents whose expertise matches the agents interests Choose the most sociable m agents of its acquaintances Choose the best m based on weighing both the expertise and the sociability

of the acquaintances Effect of Neighbor Selection Policies Choosing sociables does not help authorities to emerge 22 Decreasing Expertise; Preferring Sociables 23 Given: agents 1 and 24 lose their expertise Evolution: yet, agent 1 remains authoritative because of its sociability Increasing Expertise; Preferring Experts

24 Given: agents 79 and 237 become experts Evolution: yet, agent 79 does not become authoritative because it is a neighbor of only a few Winner Takes It All? Conjecture: After a population becomes stable, If agents prefer experts, then the winner need not take it all (i.e., a new expert can eventually become authoritative) If agents prefer sociables, then the winner takes it all (i.e., a new expert does not become authoritative)

25 Literature Referral systems: MINDS ReferralWeb Service location Directory services (WHOIS++, LDAP) No modelling of other servers Rigid referrals (if any) Chord, CAN, Pastry: Routing based on a distributed hash table No support for autonomous or

heterogeneous peers 26 Directions Practical MS Themes Reimplement MARS Incorporate QoS Research Trust model Domain ontologies Policies Virtual organizations

27 PhD Themes Ontologies An ontology is a knowledge representation of some domain of interest Successful communication (or interoperation) presupposes agreement of ontologies Currently: develop standard ontologies for each domain Time consuming; fragile Doesnt scale; omits opinions 28

IEEE SUO; Cyc; Language-based approaches: WordNet; LDOCE Consensus Referral systems are a decentralized way to achieve (or approximate) consensus About services, as above Potentially also about ontologies Use social network to determine

who is an authority in what topic Find a way to combine their ontologies for those topics 29 Big challenge: how to convincingly evaluate the contribution Logic-Based Policies Referral systems appear to work, but how can We be sure nothing bad will happen An administrator or user

configure such systems Use declarative policies to capture the agents behavior Use logic programming to develop the agents 30 Early stages: Udupi & Singh Virtual Organizations Organizations of autonomous, heterogeneous parties collaborating some computational task

Common in scientific computing Emerging in business settings Challenges VOs face Interoperation of information resources as in other systems Governance regarding allocating resources 31 Challenge to combine commitments with referral systems Key Ideas

Decentralization is desirable Leave the user in control Provide bookkeeping support Reputation in action Not separated from usage Context provides meaning to pointers Interesting properties of clustering and emergence Intuitive model underlying link analysis 32 Backup Slides 33

AutTitle Text Sidebar 34 Basic Experimental Setup Interests used to generate queries Query, answer, interest, and expertise are vectors from Vector Space Model where each dimension corresponds to a domain Dimension of the vectors is 4 Sociability is scalar 400 agents, with 10 to 25% service providers

8 neighbors per consumer Initial neighbors picked randomly Reselect neighbors after every 2 queries 4 to 20 neighbor changes 35 Metrics Qualifications: Similarity: A symmetric relation to measure how similar two vectors are Capability: An asymmetric relation to measure how much better a vector is compared to the other 36

Metrics Quality: Direct: How close a match are the neighbors of an agent to it? Nth Best: Sort them and take the highest nth value. Each agent is represented by its nth best matching neighbor PageRank: 37 Clustering Measures how similar the

neighbors of an agent are as well as how similar the agent is to its neighbors 38 Agents with similar interests May be looking for similar providers May give useful referrals Thus, will be considered sociable, and kept as neighbors Sociability increases interest clustering

Clustering (2) Result: Quality decreases when interest clustering increases 39 Co-Citation versus Referral Communities Bipartite Communities Referral Communities 40 Graph Structures

Result: In a population where each agent exercises the Providers policy, if there are more providers than the number of neighbors an agent can have, then the graph converges into a bipartite graph Bipartite Graphs Weakly-connected components Approximate how close a graph is to being bipartite: Removing k edges Removing k vertices

41 Graph Structures Result: In a population where each agent exercises the Sociables policy, the graph ends up with a number of weakly-connected components Bipartite Graphs Weakly-connected components If there is more than one weakly-connected component, then there is at least one

customer who will not be able to find a service provider 42 In-Degree Distributions Referral Policies Neighbor Selection Policies 43 Power Laws On Power-Law Relationships of the Internet Topology M. Faloutsos

P. Faloutsos C. Faloutsos (SIGCOMM 1999) Interacting Individuals Leading to Zipfs Law M. Marsili Y. Zhang (Physical Review Letters, 80(12), 1998) 44 Power-Law Distribution of In-Degree When agents are ranked based on their in-degree, the

agent with the highest rank has a lot higher in-degree than the agent with the second rank, and so on 45 Agents Prefer Providers (1) With nonselective referrals, when agents prefer providers, the in-degrees are shared among service

providers 46 Agents Prefer Sociables (1) 1. With selective referrals, agents become locally sociable 2. In-degree distribution becomes a power-law 47 Agents Prefer Sociables (2) Decreasing the selectivity of referrals decreases the fitness of the power-law 48 Discussion

Reputation? What reputation? Clearly being used in referrals Clearly being built up or torn down But not being computed as such (except for an after-the-fact study) Directions Richer representations: transfer reputation across services Protection against attacks: deception, collusion Implementation 49 Reputation Consider a society of principals,

potentially each having opinions about the others The opinions are applied implicitly in whether and how different parties do business with each other Someones reputation is a general opinion about that party Sometimes partially probed by asking others Never explicitly fully aggregated, except in current computational approaches 50

Recently Viewed Presentations

  • National Institute of Standards (NIS) Egypt Length Activities

    National Institute of Standards (NIS) Egypt Length Activities

    Range : 100 x 200 mm. uncertainty: ±5 um. You don't need to mentionaccurac. Model enough. Engineering and Surface Metrology (Laboratory 3) 3.4 -Machine tools metrology - Laser displacement interferometer. Calibration of CMM, CNC and universal microscope .
  • BEOWULF - Breal

    BEOWULF - Breal

    Characteristic features of Anglo-Saxon Literature Alliteration Kennings Alliteration The repetition of beginning consonant sounds in lines of poetry Kenning Kenning - a metaphorical phrase used to replace a concrete noun Ex: sea = whale's home Ex: battle = spear play...
  • Context Free Grammars

    Context Free Grammars

    How do we use a grammar to check the validity of data or a source code file? Bad Answer : generate. lots of possibilities until you generate a match. Then you know the data matches the grammar. Good Answer :...
  • Electric Current and Resistance

    Electric Current and Resistance

    acts as one way valve turning AC into pulsed DC. 2 or more diodes together can provide smoother DC . Capacitors also used to smooth out DC signal. Most electronic devices need DC to operate, so they must have a...
  • Fiduciary Duties of Corporate Officers

    Fiduciary Duties of Corporate Officers

    Directors' Fiduciary Duties Care: The duty of care requires that directors inform themselves of "all material information reasonably available to them," prior to making a business decision. Smith v. Van Gorkom, 488 A.2d 858, 872 (Del. 1985) (citation omitted).
  • First Grade Homework & Classroom News:Jan. 8  Jan

    First Grade Homework & Classroom News:Jan. 8 Jan

    Place value-determine how many tens and ones are in 2-digit numbers. Count forward and backwards from any number within 120. Find ten more and ten less from any number within 120. Create your own story problems using place value and...
  • Ildegarda di binsen

    Ildegarda di binsen

    Le Hildegurls sono un quartetto tutto femminile e sono riuscite a combinare i lavori di Hildegard "Ordo Virtutum" "Order of the Virtues," con quelli di quattro compositori contemporanei Nel 2006 al Festival di Santo Stefano a Bologna Claudia Koll ha...
  • powerpoint - council

    powerpoint - council

    Developing a Climate Change Action Plan for Sunderland Jim Gillon Sustainability Co-ordinator 28th November 2007 Presentation to North East Domestic Energy Forum