Skip to main content

Evolution and Ecology of Interacting Infohabitants

Objective

This project is about a study of evolution and ecology of interacting infohabitants. We will study a population of infohabitants, with carefully designed interaction. An ontology will be defined, and soft computing used as a model for interaction. Properties to be studied are scalability, openness, adaptability, and stability. We aim at deriving concrete results that can be used in the design of universal information ecosystems. A demonstrator will be constructed as well.

OBJECTIVES
Our main objective is to study the ecology of a system of interacting intelligent infohabitants, especially the scalability, openness, adaptability, and stability of the ecosystem. A sub-objective is to define intelligent interaction, or interaction of intelligent infohabitants. Another sub-objective is to study how the complexity of interaction of the infohabitants influences the behaviour of the ecology. Another sub-objective is to find what are the properties of the ecology of infohabitants that are worth studying. Another sub-objective, related to scalability, is the openness with respect to integration of new types of intelligent infohabitants. Our last technical sub-objective is the specification of conditions of stability of the ecosystem. Validation is also an objective of our project.

DESCRIPTION OF WORK
We will determine the convergence time of a system of interacting intelligent infohabitants. We will relate this time to the communication time, and the communication load. The infohabitants interact using soft computing. We will simulate this, and then derive a scaling law of the convergence time as a function of the number of infohabitants. We will grow the emergent ontology by simulation in a stable system and the environment will be partitioned in certain habitats. We will study the impact upon the ontology by the inclusion or exclusion of infohabitants, and by the extension of the behaviour of the infohabitants. Our learning infohabitants comprise:
1. The learning element
2. The performance element
3. The critic
4. The problem generator.

We plan to use fuzzy deterministic descriptions. Find what feedback is available for the learning algorithm, and what prior knowledge is available. We will use the stability of agent systems via equilibrium distributions of stochastic processes. The characteristics of these stochastic pr0cesses have to be changed, because the infohabitants are now part of an ecosystem. We will do simulations, and some theoretical work. These results then have to be adapted to the use of fuzzy variables.

Our electronic commerce process will have the following steps:
Step 1. The user enters his requirements and desired factors in a computer, using a natural language form using the shared ontology to be developed in this project
Step 2. An infohabitant is dispatched to "shop around" entering the network an communicating electronically with other infohabitants and databases to locate and obtain the required information
Step 3. The infohabitants of the previous step will attempt to match the requirements against what is available, negotiating with vendor's nfohabitants. These infohabitants might activate others to make arrangements, co-operate with each other and make inquiries
Step 4. An infohabitant returns with one or more recommendations.

Call for proposal

Data not available

Coordinator

IMPERIAL COLLEGE OF SCIENCE, TECHNOLOGY AND MEDICINE
Address
South Kensington Campus
SW7 2AZ London
United Kingdom

See on map

Participants (2)

FACULDADE DE CIENCIAS E TECNOLOGIA DA UNIVERSIDADE NOVA DE LISBOA
Portugal
Address
Ffct - Universidade Nova Lisboa, Quinta Da Torre
2829-516 Caparica

See on map

TECHNISCHE UNIVERSITEIT EINDHOVEN
Netherlands
Address
Den Dolech 2
5600 MB Eindhoven

See on map