-
The first work package WP1 concerns data collection and completion. It has three major objectives. Objective 1.1 consists in identifying data sources and collecting data/knowledge. Objective 1.2 deals with unstructured data and building terminological knowledge bases. Objective 1.3 consists in using machine learning and inference algorithms, enhanced by knowledge, for extracting and deriving missing data. The consortium has already collected (partial) data on the networks of the city of Montpellier.
-
The second work package WP2 deals with the problem of information modelling with a particular focus on defining representational languages for managing heterogeneous and uncertain information. It has two major objectives. Objective 2.1 concerns the definition languages for representing different forms of data and knowledge. Objective 2.2 aims at studying the problem of heterogeneity of data description models or uncertainty frameworks that may happen when the different information sources, to be combined, do not share a common data description language.
-
The third work package WP3 is devoted to fusion and clustering problems, with three major objectives. Objective 3.1 deals with the problem of combining, knowledge dynamic and integration. It studies different types of fusion methods that can be applied on wastewater and stormwater network data. Objective 3.2 addresses the issue of handling conflicts which is a central feature in fusion and an increasingly important topic within AI. This includes the development of techniques for measuring degrees of inconsistency and for an efficient restoring of consistency. Objective 3.3 relates to the development of classification and graph clustering methods in the context of wastewater and stormwater networks data.
-
The fourth work package WP4 has two main objectives: Objective 4.1 aims at developing tractable query answering mechanisms that take into account uncertainties and inconsistency of information/data. It also aims at proposing different strategies for explainability, including clustering solutions and argumentations, that go beyond mere weighted answers representations. It also addresses the question of the explanation according to who asks for it. Objective 4.2 concerns the
building of datasets and the development of algorithms and tools that require empirical validations where experiments will be carried out on real datasets (collected in WP1 and undergone processing steps defined in WP2-3). -
The last work package WP5 is fully dedicated to the project management, training activities and the dissemination of the research results. It covers in particular i) training sessions and communication of the research results, ii) two summer schools (one on the heterogeneity of wastewater and stormwater networks data and the other on data fusion), iii) two AI specialized workshops (one on information fusion and the other on reasoning with uncertainty), iv) a side event (for the Water Science community) and v) a final workshop aimed at a large audience.
