International Conference on Computational Collective Intelligence – Semantic Web, Social Networks & Multiagent Systems
For official information check website: ICCCI 2009.
Day 2, Tuesday, October 6, 2009.
8:30, session 1, stream A
Juan J. Pardo (speaker), Manuel Nunez, M. Carmen Ruiz: A novel formalism to represent collective intelligence in multi-agent systems. Input Output mapping system for agent actions. Based on state machines and transitions.
Robert Trypuz (speaker), Pawel Garbacz, Piotr Kulicki, Marek Lechniak: A formal model for epistemic interactions. Modal logic, AI puzzles: muddy children.
changing to stream B, Semantic Web
Kamil Karwowski, Witold Wysota, Jacek Wytrebowicz: Computer Aided Requirements Management (Project conducted for Polish Telecom, Telekomunikacja Polska). Semantic approaches and tools: OntoWiki – general tool for visualization and editing of semantic knowledge base. SoftWiki – tool for aiding software development, SEOntology (software engineering ontology). Contains soft.eng and project management ontology, together with domain specific layer. System CARM: Thin-client through a web-browser. Ability to work offline (after downloading the data to a local storage). Following USA DoD Architectural Framework (DoDAF), and UK Ministry of Defense AF (MoDAF) architectural recommendations. JENA ontology engine, Pellet reasoner, Tomcat for deployment mgmt.
Adam Meissner (Poznan University of Technology): A simple parallel reasoning system for the alc description logic. The use of Oz multiparadigm programming environment. Running tests with coordinator and up to five workers on a small cluster of workstations with Windows OS and Mozart programming environment. See the paper for results.
10:45 – Session 2, Stream A
Drago Zagar, Slavko Rucic and Snjezana Rimac-Drlje (Environment University, Croatia): Agent based architecture for Pricing the services in Dynamic User/Network. The studey of dynamic pricing and QoS-enabled pricing. Problem with estimation and verification of Quality of Service in the context of accounting (rating) network access. Classical pricing schemes are outdated: (a) the flat schemes do not stimulate improvement in network; (b) future networks have to deal with numbers of apps with very diverse QoS. New, more dynamic pricing schemes are necessary. Various levels of QoS (Integrated Services and Diff.Services). The problem is the perceived QoS, the delivered QoS and the value or price for a given QoS. Presentation is about the model. Future work: formal specification, verification, implementation, and building the MAS.
Rustam Tagiew (Technische Universitat, Fraiberg, Germany, originally from Azarbejdzan): On Multi-Agent Petri net models for Computing Extensive Finite Games. Petri-nets for Strategic Interactions (PNSI). PNSI offers: concurrency and graphical representation. Petri nets (with weighted transitions) are extended by the notion of agents. The game is a strategic interaction between two agents. Description of the classes of two-player games: zero-sum, deterministic, perfect information, imperfect information (not everyone can see every-actions), incompete information (some players do not know the rules). Review of existing game description languages:
- Gala = GAme LAnguage (prolog-based), games of imperfect information no notion of time, 1997 Pfeffer and Koller
- GDL = Game Description Language Genesereth et al 2005,
- TPN = Time Petri Nets Merlin 1974 and Timed Petri Nets Ramchandi (1974).
- GAMBIT – state of art library for solving games. Representation in strategic and extensive form.
Ali Selamat, Muhammad Tarmizi Lockman (Technical University of Malaysia, www.utm.my): Multi-agent Verification of RFID System. Supply chain mngmt asset tracking, identification of items. Multi-tag reading. Problems of decoding and verification.
David Vallejo Fenandez:A multi-agent architecture for multi-robot survaillance, University of Castilla, Spain, ORETO (UCLM) – RoboKU (Kingston Iniversity). Context: evolution of surveillance systems, driven by AI techniques, intelligent component, distributed components, high-number of devices. Specific domain: multi-robot coordination. Coordination same as in MAS, but extra elements: uncertainty, management, physical communication, highly dynamic environments. Describing the architecture: perceptual layer, conceptual layer together with reasoning core, and the final element for decision making. See the article for implementation details.
13:45 Planary session: Pierre Levy, University of Ottawa.
Toward a Civilization of Collective Intelligence Royal Society of Canada. Twitter: @plevy
People as symbol manipulators, placed in a social context. Collective symbol manipulators. Cyberspace (scientific observatory/digital mirror of Collective Intelligence) <-> Reflexive Collective Intelligence (driver of human development) <-> Human Development (prosperity, health, education, security, peace, research, heritage, innovation). The target: 2015, Semantic Space, Interconnection between ideas (via semantic tags). Uniform Semantic Locator = IEML * concept address. Collaborative societies of semantic agents, subject-centric computation. Collective intelligence growth. Augmentation of sense-making. The new language, IEML, has the following primitives: emptiness, virtual, actual, signifier (sign), referant (the thing), signified (concept, the being) (see Aristotle for the last 3. Tertiary trilectic).
15:00 Panel Discussion “Is collective better than individual”?
Ngoc Thanh Nguyen:
- When a collective is more intelligent than a single?
- How to determine the intelligence of a collective on the bassi of its members’ intelligence?
- Is the inconsistency among participants useful for collective intelligence?
Ryszard Kowalczyk: defines collective intelligence as a behaviour (of a system) that emerges from interaction of many autonomous individuals. Two perspectives: individuals&interactions -> system, system -> individuals&interactions, “sometimes it is better to know less for the system to exhibit optimal performance and efficiency as a whole”.
Piotr Jedrzejowicz: What CI is trying to address?
- no free lunch theorem
- complexity and uncertainty
- constructing cognitive machines
Roman Slowinski: the problem of aggregation, aggregation of evaluations.
Day 3, Wednesday, October 7, 2009.
Session 1, Stream A (I am the chair)
Aziz Aziz, Jan Treur: Modelling Dynamics of Social Support Networks for Mutual Support in Coping with Stress
Session 2, Stream A
Krzysztof Kulakowski: The rorma game on a model network: a critical line “The first purpose of any social system is to create incenties for individuals to conduct more work then the one requiredby their immediate wants (needs)”.
Wojciech Kaminski, Marek Chlebus, Romuald Kotowski (Polish -Japanese Institue of Information Technology): Collective Prisoner’s Dilemma Model of Artificial Society. Prisoner’s dilemma model with mood, emotions, opinions, etc. Agents interact each time with a different agent. Modelling personality traits within individual agents: – rationality: how much emotions influence the decision; – sensitivity: how much actions influence emotions; – openness: the will to interact with strangers; – tolerance: – activity: – conformism: public opinion influence; – morality model: commercial morality, utilitarian morality; – cooperative morality over 2700 different personalities. Agent’s current state consists of: emotions, reputation, memory, economic strength.
A surprising result with only IncomeTax enabled – the economy folds down for many generations, and then boosts suddenly and growth exponentially.