Research regards Cooperative Scientific Environments for distributed experiments. It aims at combining Large-scale collaboration techniques in today scientific experiments, involving a variety of digital resources, and Cooperative Information Systems.

Research regards:

· Modeling scientific experiments as distributed processes (workflows or BP);

· Management of networks of cooperative e-nodes;

· Trust and reputation in cooperative environments;

· Open Source Software for bioinformatics experiments;

· Agent and Web Service-based architectures for coordinating experiments.

Note:

The concept of “what an experiment is” is rapidly changing in an ICT oriented environment, moving from the idea of a local laboratory activity towards a computer and network supported application including the integration of:

· a variety of information and data sources;

· the interaction with physical devices;

· the use of heterogeneous software systems.

In our approach, scientific experiments are modeled analogously to cooperative enterprise processes, as e-processes that operate on, and manipulate, data sources and physical devices.

All services have an e-node responsible for offering the service and which sets the rules under which the service can be accessed by other e-nodes through service invocation.

An experiment involves multiple e-nodes interacting with one another in order to offer or to ask for services.

The user is a researcher who has the following possibilities:

· selection of the experiment of interest and of the information sources he/she wants the experiment be carried on;

· acquisition and collection of local data;

· surveillance/monitoring of local experiments, which are part of a cooperative experiment;

· definition of new experiments; this implies that the global workflow of the experiment must be designed;

· inspection of remote data sources and experiment results, e.g., by mining in a data warehouses;

· cooperation with users of other e-nodes, for example to co-design experiments and jointly evaluate results.

· A Pool of Services (PS) is an e-node described by:

· A set of local information sources;

· A set of local services.

Four main classes of experiments have been tested:

· Process Simulation and Visualization.

· Supervised or Unsupervised Classification.

· Machine Learning: Rule Generation and Bayesian Networks

· Identification of the Process Dynamics.

Implementation

A first developed suite of prototypes is available at http://www.dsf.unica.it/~andrea/webservices.html.

Cooperation is active with Consiglio Nazionale delle Ricerche-Milano, and Università degli Studi di Cagliari, Dipartimento di Matematica e Informatica.