Billing the Grid – Kick Off Meeting

Slides:



Advertisements
Ähnliche Präsentationen
Service Oriented Architectures for Remote Instrumentation
Advertisements

PRESENTATION HEADLINE
An new European Power Network: Student Power
PSI and Competition The General Framework
Finding the Pattern You Need: The Design Pattern Intent Ontology
E-Solutions mySchoeller.com for Felix Schoeller Imaging
H - A - M - L - E - IC T Teachers Acting Patterns while Teaching with New Media in the Subjects German, Mathematics and Computer Science Prof. S. Blömeke,
Managing the Transition from School-to-Work Empirical Findings from a Mentoring Programme in Germany Prof. i.V. Dr. Martin Lang.
R. Zankl – Ch. Oelschlegel – M. Schüler – M. Karg – H. Obermayer R. Gottanka – F. Rösch – P. Keidler – A. Spangler th Expert Meeting Business.
Herzlich Willkommen zum Informations-Forum: SAP Interoperabilität
The difference between kein and nicht.
© 2006 Open Grid Forum OGF26 - Chapel Hill, May 2009 Addressing Metadata Challenges OGF Digital Repositories RG.
First Seminar in Brussels, 15th of December 2010
Institut für Umweltschutz in der Berufsbildung e.V. Folie 1 ALBATROS Ziele Jahreskonferenz 2004 Ziele der Albatros-Jahreskonferenz 2004 Objectives of the.
Die Senatorin für Arbeit, Frauen, Gesundheit, Jugend und Soziales ESF-Verwaltungsbehörde Freie Hansestadt Bremen Hildegard Jansen, head of Unit labour.
Steinbeis Forschungsinstitut für solare und zukunftsfähige thermische Energiesysteme Nobelstr. 15 D Stuttgart WP 4 Developing SEC.
Fakultät für informatik informatik 12 technische universität dortmund Optimizations Peter Marwedel TU Dortmund Informatik 12 Germany 2009/01/17 Graphics:
Fakultät für informatik informatik 12 technische universität dortmund Specifications Peter Marwedel TU Dortmund, Informatik 12 Graphics: © Alexandra Nolte,
Peter Marwedel TU Dortmund, Informatik 12
Fakultät für informatik informatik 12 technische universität dortmund Hardware/Software Partitioning Peter Marwedel Informatik 12 TU Dortmund Germany Chapter.
Telling Time in German Deutsch 1 Part 1 Time in German There are two ways to tell time in German. There are two ways to tell time in German. Standard.
Rethinking Linguistic Relativity John A. Lucy. Gliederung Einführung in das Problem Kritik an bisherigen Untersuchungen der Anthropologen Psycholinguisten.
Regional Support in the context of LCG/EGEE
NUMEX – Numerical experiments for the GME Fachhochschule Bonn-Rhein-Sieg Wolfgang Joppich PFTOOL - Precipitation forecast toolbox Semi-Lagrangian Mass-Integrating.
Institut für Verkehrsführung und Fahrzeugsteuerung > Technologien aus Luft- und Raumfahrt für Straße und Schiene Automatic Maneuver Recognition in the.
Institut für Verkehrsführung und Fahrzeugsteuerung > Technologien aus Luft- und Raumfahrt für Straße und Schiene Driving Manoeuvre Recognition > 19. Januar.
Lancing: What is the future? Lutz Heinemann Profil Institute for Clinical Research, San Diego, US Profil Institut für Stoffwechselforschung, Neuss Science.
SOLNET Milan Optimization of the air-to-water heat exchanger configuration for water preheating in open district heating nets Janybek Orozaliev.
Lehrstuhl Informatik III: Datenbanksysteme AstroGrid-D Meeting Heidelberg, Informationsfusion und -Integrität: Grid-Erweiterungen zum Datenmanagement.
Hochschulteam der Agentur für Arbeit Trier Preventing the Brainware Crisis Workshop Schloss Dagstuhl Student Enrollment in Computer Science.
5th IAEA Technical Meeting on ECRH Gandhinagar – February 2009
Introduction to the topic. Goals: Improving the students essay style in general Finding special words and expressions that can be used in essay writing.
Fachabteilung 16A Überörtliche Raumplanung Cross border co-operation from the view of a public administration unit.
Deutsche Gesellschaft für Technische Zusammenarbeit GmbH Integrated Experts as interface between technical cooperation and the private sector – An Example.
Seminar Telematiksysteme für Fernwartung und Ferndiagnose Basic Concepts in Control Theory MSc. Lei Ma 22 April, 2004.
Methods Fuzzy- Logic enables the modeling of rule based knowledge by the use of fuzzy criteria instead of exact measurement values or threshold values.
Comparative Adjectives. The term comparison of adjectives is used when two or more persons or things have the same quality (height, size, color, any characteristic)
Institut für Umweltphysik/Fernerkundung Physik/Elektrotechnik Fachbereich 1 SADDU June 2008 S. Noël, K.Bramstedt,
20:00.
Laurie Clarcq The purpose of language, used in communication, is to create a picture in the mind and/or the heart of another.
Machen Sie sich schlau am Beispiel Schizophrenie.
Die Zeit (TIME) Germans are on military time which is 1-24
Institut AIFB, Universität Karlsruhe (TH) Forschungsuniversität gegründet 1825 Towards Automatic Composition of Processes based on Semantic.
Sanjay Patil Standards Architect – SAP AG April 2008
| DC-IAP/SVC3 | © Bosch Rexroth Pneumatics GmbH This document, as well as the data, specifications and other information set forth in.
BAS5SE | Fachhochschule Hagenberg | Daniel Khan | S SPR5 MVC Plugin Development SPR6P.
Analysis of Cross-Polarization Modulation in Dispersion-Managed DWDM Systems Marcus Winter, Christian-Alexander Bunge, Dario Setti, Klaus Petermann LEOS.
1 Ein kurzer Sprung in die tiefe Vergangenheit der Erde.
Alp-Water-Scarce Water Management Strategies against Water Scarcity in the Alps 4 th General Meeting Cambery, 21 st September 2010 Water Scarcity Warning.
INTAKT- Interkulturelle Berufsfelderkundungen als ausbildungsbezogene Lerneinheiten in berufsqualifizierenden Auslandspraktika DE/10/LLP-LdV/TOI/
Einführung Bild und Erkenntnis Einige Probleme Fazit Eberhard Karls Universität Tübingen Philosophische Fakultät Institut für Medienwissenschaft Epistemic.
Berner Fachhochschule Hochschule für Agrar-, Forst- und Lebensmittelwissenschaften HAFL Recent activities on ammonia emissions: Emission inventory Rindvieh.
4th Symposium on Lidar Atmospheric Applications
Ein Projekt des Technischen Jugendfreizeit- und Bildungsvereins (tjfbv) e.V. kommunizieren.de Blended Learning for people with disabilities.
Talking about yourself
Cross-Polarization Modulation in DWDM Systems
ESSnet Workshop Conclusions.
1 von 10 ViS:AT Abteilung IT/3, IT – Systeme für Unterrichtszwecke ViS:AT Österreichische Bildung auf Europaniveau BM:UKK Apple.
Berner Fachhochschule Hochschule für Agrar-, Forst- und Lebensmittelwissenschaften HAFL 95% der Ammoniakemissionen aus der Landwirtschaft Rindvieh Pflanzenbau.
Wind Energy in Germany 2004 Ralf Christmann, BMU Joachim Kutscher, PTJ
KIT – die Kooperation von Forschungszentrum Karlsruhe GmbH und Universität Karlsruhe (TH) Vorlesung Knowledge Discovery - Institut AIFB Tempus fugit Towards.
1 Stevens Direct Scaling Methods and the Uniqueness Problem: Empirical Evaluation of an Axiom fundamental to Interval Scale Level.
Technische Universität München Spatial aspects of the formation of GMO-free or GMO clubs Maarten J. Punt Technische Universität München.
Lehrstuhl für Waldbau, Technische Universität MünchenBudapest, 10./11. December 2006 WP 1 Status (TUM) Bernhard Felbermeier.
Selectivity in the German Mobility Panel Tobias Kuhnimhof Institute for Transport Studies, University of Karlsruhe Paris, May 20th, 2005.
How to use and facilitate an OptionFinder Audience Response System.
Technische Universität München 1 CADUI' June FUNDP Namur G B I The FUSE-System: an Integrated User Interface Design Environment Frank Lonczewski.
TUM in CrossGrid Role and Contribution Fakultät für Informatik der Technischen Universität München Informatik X: Rechnertechnik und Rechnerorganisation.
Andreas Burger ZENTRUM FÜR MEDIZINISCHE LEHRE RUHR-UNIVERSITÄT BOCHUM Irkutsk October 2012 Report about the lecture "Report of the TEMPUS IV- Project Nr.
Inter-Cultural Teaching and Learning ICTaL Technische Universität Berlin Zentraleinrichtung Kooperation Wissenschaftliche und interne Weiterbildung Introductory.
 Präsentation transkript:

Billing the Grid – Kick Off Meeting

Begrüßung & Vorstellung Agenda Uhrzeit Thema Zuständigkeit 11.00-11.15 Begrüßung & Vorstellung Christof Weinhardt 11.15-11.30 Vortrag Günter Quast 11.30-11.55 Christian v.d. Weth 11.55-12.10 Arun Anandasivam 12.10-12.50 Zielsetzung und Organisation: Weichenstellung Erste Schritte Reporting / Meetings Interne Verrechnung Alle 12.50-13.00 Fazit

Begrüßung & Vorstellung Agenda Uhrzeit Thema Zuständigkeit 11.00-11.15 Begrüßung & Vorstellung Christof Weinhardt 11.15-11.30 Vortrag Günter Quast 11.30-11.55 Christian v.d. Weth 11.55-12.10 Arun Anandasivam 12.10-12.50 Zielsetzung und Organisation: Weichenstellung Erste Schritte Reporting / Meetings Interne Verrechnung Alle 12.50-13.00 Fazit

A Unifying Framework for Behavior-based Trust Models Christian von der Weth, Klemens Böhm Universität Karlsruhe (TH), Germany {weth|boehm}@ipd.uni-karlsruhe.de

Motivation Many fields of research require resource-intensive applications (analysis, simulation, visualization, etc.) Real driving force: Particle Physics Solution: Grid Computing Participants (institutes, firms, persons, etc.) provide their own resources and share them with others A participant can interact with partners to use their resources to run his own applications Characteristic of Grid communities Participants have full control over their entities  A partner can impair the outcome of an interaction by behaving uncooperatively, maliciously or defectively (close access to his resources, limit bandwidth/CPU/…) Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Motivation Goal: Mechanism that allows entities autonomously to distinguish good from bad partners Promising approach: Behavior-based trust Trust: "One's subjective degree of belief that a partner can and will perform a specific task in a certain situation." Behavior-based: The trust in a partner is derived from the knowledge about his behavior in previous interactions Basic Idea: Enabling users to define their own policies whether a partner is trustworthy or not ( trust policies) and Making these policies explicit to their controlled entities Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Behavior-based Trust Policies Example policies: Alice: "I deem a partner trustworthy to use my resources if the average feedback value about him is positive." Bob: "A partner can have 100% of my idle CPU time if there is no negative feedback about him within the last 24h." Carol: "I only perform the task of others if their performance of complex tasks was satisfactorily." Dave: "A partner can have limitless bandwidth if the k most reputable entities recommend him." Eve: "I share my resources only with the k entities that have the highest PageRank." Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

What can we learn from the examples? Requirement 1: Representation of knowledge that describes the behavior of a partner: behavior-specific knowledge Different types of behavior-specific knowledge  Feedback, Reputation, Recommendation, Trust Consideration of various aspects of the behavior-specific knowledge (e.g., context, age of knowledge, etc.) Requirement 2: Mechanism makes trust policies explicit to controlled entities Different user have different trust policies Trust policies may require complex operations (e.g., aggregation or centrality computation) Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

What can we learn from the examples? (2) Representation of knowledge as directed graph G(V,E) V…set of participants E…set of edges based on behavior-specific knowledge Example:  Application of graph algorithms to find trustworthy partners e.g., EigenTrust (Schlosser et al., 2003), PageRank (Brin and Page, 1996) A Feedback Recommendation Trust B C E D Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Status Quo Existing behavior-based trust models Definition of the representation of behavior-based knowledge Definition of a fixed evaluation scheme to derive the trust in a partner  A fixed evaluation scheme contradicts the subjective nature of trust Common approach for making trust policies explicit: Logic-based trust policy languages Definition of rules and clauses to derive the trustworthiness of a partner  Existing languages cannot satisfactorily cope with complex operations required by various behavior-based policies So…what does current literature say to this problems? Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

A Framework for behavior-based trust models Aspects of our framework Relational representation of behavior-specific knowledge Algebra-based language for the formulation of behavior-based trust policies Advantages Supports the definition of arbitrary user-defined trust policies for behavior-based trust models Including all existing evaluation schemes from literature we are currently aware of Relational representation allows for a straightforward implementation Instead of defining a trust model with a fixed evalutaion scheme, we propose a framework for behavior-based trust models Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Agenda Introduction Representation of behavior-based knowledge Definition of a query algebra for trust Preliminary Performance Experiments Summary & Outlook Introduction Knowledge Representation A Query Algebra for Trust Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Types of behavior-specific knowledge (1) Feedback An entity's (rater) rating of an interaction performed by a partner (ratee) Alice: "The last download from Bob was very reliable." Recommendation An entity's (recommender) opinion about the previous behavior of a partner (recommendee) Alice: "For downloads I can recommend Bob." Introduction Knowledge Representation Overview Aspects Relational Representation A Query Algebra for Trust Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Types of behavior-specific knowledge (2) Reputation General opinion of the whole network towards a single entity  Global characteristic of an entity Example: "With regards to downloads, Bob has an excellent reputation." Trust An entity's (truster) degree of belief that a partner (trustee) will behave as expected Alice: "I trust Bob regarding the provision of reliable downloads." Introduction Knowledge Representation Overview Aspects Relational Representation A Query Algebra for Trust Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Aspects of Behavior-specific Knowledge (1) Value ∈ [-1,1] Continuous valuation allows for a finer granularity Alice: "The performance of Bobs last computation was quite good (~0.6)." Context Allows to distinguish between different situations in which two entities can interact Alice: "Bob provided fast downloads but his CPU performance was very poor." Facets of a context Allows to distinguish between different perspectives of a context Alice: "The connection for the last download was very stable but unfortunately very slow." Introduction Knowledge Representation Overview Aspects Relational Representation A Query Algebra for Trust Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Aspects of Behavior-specific Knowledge (2) Timestamp Allows to emphasize the impact of current knowledge Alice: "Bobs early downloads were quite fast but recent ones were very slow." Certainty ∈ [0,1] Allows to quantify the certainty of an assessment Alice: "I am absolutely sure (e.g., ~1.0) that Bobs performance according to his last computation was good." Estimated Effort ∈ [0,1] Allows to quantify the perceived complexity of an interaction Alice: "Bob performed simple (e.g., ~0.2) computations quite good but complex ones (e.g., ~0.9) very poor." Introduction Knowledge Representation Overview Aspects Relational Representation A Query Algebra for Trust Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Relational Representation of Knowledge Relations that represent behavior-specific knowledge: Feedback, Recommendation, Reputation, Trust (Additional relation: Entity(ID) Alice: "I am quite sure that the download from Bob was very fast. It was a big file."  New Feedback tuple In our scenario: Only Feedback tuples reflect direct experiences Other knowledge must be derived from feedback (including Trust tuples)  Goal: Trust policy language as mechanism to derive Trust, Recommendation and Reputation tuples Introduction Knowledge Representation Overview Aspects Relational Representation A Query Algebra for Trust Experiments Summary&Outlook Rater Ratee Value Context Facet Time Certainty Effort Alice Bob 0.95 Download Speed 12:09:45 0.75 0.8 Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Approach to an Algebra-based Policy Language Source: Relational representation of knowledge Evaluation of a trust policy = Query on the knowledge base Common way to deal with relations: Relational Algebra (RA) Set of operators for the application on relations Closure property of the operators allows for nesting of the operators to more complex algebra expressions  Basic Idea: Relational Algebra (RA) as basis for our trust policy language Introduction Knowledge Representation A Query Algebra for Trust Basic Idea Conventional Extensions Centrality Operator Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Example Trust Policy Informal formulation: "I trust you (idpartner) in context c and facet fc if your average feedback value from the 10 most reputable entities tops a specific threshold." Only feedback tuples with a certainty>0.8 should be considered  Algebra expression of that policy: Introduction Knowledge Representation A Query Algebra for Trust Basic Idea Conventional Extensions Centrality Operator Experiments Summary&Outlook PROJECTION[trusted]( MAP[trusted, (avg_value>threshold)]( GROUP[avg_value, AVG(Feedback.value), {ratee}]( JOIN[Feedback.rater=Reputation.entity]( TOP[10, Reputation.value]( SELECTION[context=c, facet=fc](Reputation) SELECTION[ratee=idpartner, context=c, facet=fc, certainty>0.8](Feedback) ) ) ); Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Algebra-based Policy Language Observation: Basic operators of the RA are not sufficient for the formulation of behavior-based trust policies Extension by means of additional operators are necessary  Clarification which further operators are essential to provide the desired expressiveness First step: Existing additional operators from literature Top operator (e.g., Bertino et al., 2004) Map operator (e.g., Aberer and Fischer, 1995) Introduction Knowledge Representation A Query Algebra for Trust Basic Idea Conventional Extensions Centrality Operator Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Conventional Extensions to the RA (1) Top Operator: TOP[k,attr](relation) returns the k tuples with the highest value of a attribute attr Example: Introduction Knowledge Representation A Query Algebra for Trust Basic Idea Conventional Extensions Centrality Operator Experiments Summary&Outlook ID … Value Bob 0.71 Carol 0.95 Alice 0.98 Eve 0.75 Dave 0.90 TOP[3, Value](Reputation) ID … Value Carol 0.95 Alice 0.98 Dave 0.90 Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Conventional Extensions to the RA (2) Map Operator: MAP[attr,expression(A1,...,An)](relation) Allows the execution of user-defined functions over the attributes of a relation The functions are separately applied to each single tuple of the relation; the results are stored as a new attribute Example: Introduction Knowledge Representation A Query Algebra for Trust Basic Idea Conventional Extensions Centrality Operator Experiments Summary&Outlook Rater Ratee … Value Effort Alice Bob 1.0 0.2 Carol 0.8 0.9 MAP[Weighted, (Value*Effort)](Feedback) Rater Ratee … Value Effort Weighted Alice Bob 1.0 0.2 Carol 0.8 0.9 0.72 Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Centrality Indices Centrality index Example: Graph-based measure to quantify the importance of a vertex according to the graph structure Different existing measures: Indegree, PageRank, Proximity Prestige, HITS, Integration & Radiality, etc. Different measures yield different rankings Example: Introduction Knowledge Representation A Query Algebra for Trust Basic Idea Conventional Extensions Centrality Operator Experiments Summary&Outlook A Indegree PageRank A 2.0 0.23 B 0.6 0.21 C 1.8 0.31 D 0.7 0.15 E 0.3 0.1 1.0 0.9 B 0.6 1.0 0.2 C 0.2 0.5 0.1 0.9 E D Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

An Operator for Centrality Computation Requirements for a centrality operator: Flexible specification of the underlying graph  e.g., choice of the weight of an edge: "Value" vs. "Weighted" Support of various centrality measures within one operator  Definition of centrality operator: CENTRALITY[attr, Av, As, At, Aw, Measure](Rvertices, Redges) Introduction Knowledge Representation A Query Algebra for Trust Basic Idea Conventional Extensions Centrality Operator Experiments Summary&Outlook Rater Ratee … Value Effort Weighted Alice Bob 1.0 0.2 Carol 0.8 0.9 0.72 Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Centrality Operator - Example Recommendation Entity A Recommender Recommendee … Value A C 0.9 E 0.2 B 1.0 D 0.5 0.6 ID A B C D E 1.0 Introduction Knowledge Representation A Query Algebra for Trust Basic Idea Conventional Extensions Centrality Operator Experiments Summary&Outlook 0.9 B 0.6 1.0 0.2 C 0.2 0.5 0.1 0.9 E D CENTRALITY[PageRank, ID, Recommender, Recommendee, Value, PageRank] (Entity, Recommendation) ID PageRank A 0.23 B 0.21 C 0.31 D 0.15 E 0.1 Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Centrality Operator Nature of centrality computation Very time-consuming and resource-intensive  Centrality computation is the most costly part of the evaluation of a trust policy Implemented centrality measures in PL/SQL (Oracle 10g) PageRank, Positional Power Function (eigenvector centrality measures based on power iteration implementation) Authorities, Proximity Prestige, Integration Experiments Efficiency: Performance of our implementations Quality of Centrality Measures: Comparison of ranking results Introduction Knowledge Representation A Query Algebra for Trust Basic Idea Conventional Extensions Centrality Operator Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Efficiency (1) Setup: Measured value: time in sec Result All centrality measures Network sizes: 500, 1000, 2000 entities Measured value: time in sec Result Performance varies significantly from measure to measure Eigenvector centrality measures (based on power iteration implementation) show best performances Introduction Knowledge Representation A Query Algebra for Trust Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Efficiency (2) Setup: Measured value: time in sec Result: Eigenvector centrality measures Network sizes: 2000, 10000, 50000, 100000 entities Measured value: time in sec Result: Again, huge difference between both measures Main factor: error threshold of power iteration implementation (causes the number of iteration steps) Introduction Knowledge Representation A Query Algebra for Trust Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Quality of Centrality Measures Setup: All centrality measures Network size: 1000 entities Measured value: Difference between two rankings in % Mean distance between the position of an entity in both rankings 0%...equal rankings, 100%...maximum difference Result: Most measurements yield different rankings (except for Integration and Proximity Prestige) Choice of centrality measure might influence the result of trust policies significantly Introduction Knowledge Representation A Query Algebra for Trust Experiments Summary&Outlook PWF Authorities PPrestige Integration PageRank 6.2% 8.2% 5.3% - 5.4% 9.5% 9.7% 0.0% Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Summary What have we done so far? Collection of various meaningful behavior-based trust policies from literature and our own attempts Motivation of an algebraic approach for the formulation of behavior-based trust policies Definition of a relational representation of behavior-specific knowledge Definition of a query algebra for trust Listing of necessary operators from literature (basic operators from the RA incl. existing extensions) Definition of a centrality operator for the computation of various centrality measures Presentation of some first experimental results Introduction Knowledge Representation A Query Algebra for Trust Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Open Questions How efficient is the evaluation of various trust policies? Further efficiency test including various optimization techniques for centrality computation Evaluation of trust policies in distributed architectures (i.e., structured Peer-to-Peer systems) How about effectiveness when entities with different trust policies interact repeatedly? Introduction Knowledge Representation A Query Algebra for Trust Experiments Summary&Outlook Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Thanks for your interest! Questions? Christian von der Weth, Klemens Böhm: "A Unifying Framework for Behavior-based Trust Models"

Begrüßung & Vorstellung Agenda Uhrzeit Thema Zuständigkeit 11.00-11.15 Begrüßung & Vorstellung Christof Weinhardt 11.15-11.30 Vortrag Günter Quast 11.30-11.55 Christian v.d. Weth 11.55-12.10 Arun Anandasivam 12.10-12.50 Zielsetzung und Organisation: Weichenstellung Erste Schritte Reporting / Meetings Interne Verrechnung Alle 12.50-13.00 Fazit

Virtuelle Währungen als Anreizmechanismus für Grids

Reputationsmechanismen Beispiel für Reputationsmechanismus: eBay Mechanismen für P2P: EigenTrust, PeerTrust, DMRep Ziel: bösartiges und egoistisches Verhalten minimieren Mehr Vertrauen des Käufers in Händler mit guter Reputation Anreiz für Teilnehmer: Verbesserung der eigenen Reputation und folglich mehr Umsatz Nachteile: Erfüllung der Mindestanforderung ausreichend Kollusion White washing

Monetäre Mechanismen Leistung ↔ Gegenleistung in Geld Beschränkung und Kontrolle des Gesamtbudgets im System notwendig Anreiz für Teilnehmer: Leistung anbieten → Geld verdienen → Leistung erhalten Preis spiegelt Knappheit wider Nachteile: Befürchtung im universitären Bereich: Bessere Ausgangssituation für finanziell gut ausgestattete Institute.

Stamp Trading [Nach Moreton und Twigg 2003] Stamp Trading (nach Moreton & Twigg) Jeder Nutzer in Besitz seiner eigenen, persönlichen Marken Gleicher Wert für alle Marken (z.B. nur 10€ Scheine) Zahlung: Handel zwischen Person X und Person Y nur möglich mit Marken Reputation: Abhängigkeit des Markenwertes von der Anzahl der Einlösung und der Erfüllung der nachgefragten Leistung Regelung des Markenwertes durch eine zentrale Instanz für Wechselkurse Bestimmung des Markenwerts durch eine geeignete anreizkompatible Funktion, Bsp: w = m * rs / i Reputationsmechanismen Monetäre Mechanismen

Verteilung der Marken

Entwicklung eines dezentralen Ansatzes für Stamp Trading Ausblick Vorteile: Rückverfolgbarkeit möglich (Dokumentation der Zahlungsflüsse durch zentrale Instanz) Reputation und Zahlung in einem System (Marken) erfasst Nachteile: Zentrale Verwaltung der Wechselkurse notwendig  Nachteil der Skalierbarkeit Profilerstellung über die Nutzer durch zentrale Verwaltung. Systemabsturz durch technische (und juristische) Attacken auf die zentrale Einheit Eingelöste Marken nicht automatisch durch die andere Partei gelöscht  mehrmaliges Benutzen einer Marke (Double spending) Kollusionen und White washing möglich Entwicklung eines dezentralen Ansatzes für Stamp Trading

Begrüßung & Vorstellung Agenda Uhrzeit Thema Zuständigkeit 11.00-11.15 Begrüßung & Vorstellung Christof Weinhardt 11.15-11.30 Vortrag Günter Quast 11.30-11.55 Christian v.d. Weth 11.55-12.10 Arun Anandasivam 12.10-12.50 Zielsetzung und Organisation: Weichenstellung Erste Schritte Reporting / Meetings Interne Verrechnung Alle 12.50-13.00 Fazit

Mitarbeiterstruktur EKP AIFB Christian v. d. Weth Arun Anandasivam IPD ??? EKP AIFB A. Ankolekar Integration in AIFB durch Besuch der Oberseminare Integration durch … Christian v. d. Weth Arun Anandasivam IPD IISM ??? D. Neumann

Einordnung der Billing Dienste Grid Applikation Billing Dienst 2 (Virtuelle Währungen) Billing Dienst 1 (Reputationsmechanismus) Common Virtualization Middleware (Globus GT4)

Entwurf und Realisierung einer anreizkompatiblen Billing-Infrastruktur Zielsetzung Projektziel: Entwurf und Realisierung einer anreizkompatiblen Billing-Infrastruktur Praxis Theorie Anforderungsanalyse für Mechanismen Integration des Prototyps in bestehende Grid Middleware Feldexperiment Evaluation Konzeption eines Billing-Mechanismus Reputationsmechanismus Virtuelle Währung … Konzeption eines „Policy-basierte Bewertungsautomaten“  Anforderung an Infrastruktur Dezentral strukturierte P2P-Technologie für eine koordinatorfreie Datenhaltung und hohe Skalierbarkeit

Billing the Grid und KIT Adaption und Veränderung Vorhandene Schnittstellen? Reputations- mechanismen RZ FZK (Mickel) RZ Karlsruhe (Juling) Cluster Teilchenphysik CERN? D-Grid Integrationsprojekt Institut X Zeit Ansprechpartner? Pilotprojekt?

Meilensteine Meilenstein 1 Meilenstein 2 Meilenstein 3 Meilenstein 4 Anforderungserhebung Literaturrecherche Erster Prototyp Erste Ergebnisse Alternative Ansätze Feldexperiment Verbesserter Prototyp Berichte Folgeantrag 04.08.2006 01.02.2007 01.08.2007 01.02.2008 01.08.2008 Phase „Vorbereitung“ Phase „Forschung und Entwicklung“ Phase „Evaluation“

First steps (1/2) Anforderungsanalyse für Anreizmechanismen (AP10) : Domänenstrukturierung Erhebung Anreizprobleme Bösartiges vs. egoistisches Verhalten Identifikation Wissensressourcen Ableitung Anforderungen an Anreizmechanismus Ziele Lösung der Anreizprobleme Performanz  Usability/Sicherheit … Funktionale Anforderung Prozessablauf Interaktion mit dem Benutzer Grenzen vorhandener Anreizmechanismen D-Grid Integrationsprojekt SORMA Definition geeigneter Metriken

First steps (2/2) P2P Netzwerk (AP1) Konzeption eines strukturierten P2P Netzwerkes Content Adressable Network Speicherung von Feedback und anderen Metadaten Implementierung eines strukturierten P2P Netzwerkes Roll-Out

Organisation Reports Buchung Meetings PR Regelmäßigkeit der internen Reports Externer Report (Abschlussbericht) Reports Buchung Institutsintern oder institutsübergreifend? Intervalle / Zeitpunkte Treffen aller Beteiligten (2x im Jahr?) Kleine Treffen (1x pro Woche bzw. Monat) Meetings Inhalt der Homepage (www.billing-the-grid.org) Logo PR

Anschubfinanzierung „Landesschwerpunktprogramm erwartet Antragstellung“ BMBF EU-Projekt FP7 IST DFG SPP DFG Forschergruppe Welches Ziel wird nach dem Projekt verfolgt? Ist ein Folgeprojekt erforderlich? Sorma EU-Projekt FP6 Call5 Biz2Grid …

Begrüßung & Vorstellung Agenda Uhrzeit Thema Zuständigkeit 11.00-11.15 Begrüßung & Vorstellung Christof Weinhardt 11.15-11.30 Vortrag Günter Quast 11.30-11.55 Christian v.d. Weth 11.55-12.10 Arun Anandasivam 12.10-12.50 Zielsetzung und Organisation: Weichenstellung Erste Schritte Reporting / Meetings Interne Verrechnung Alle 12.50-13.00 Fazit