Research Project - EICOS

How do we escape from the rigidity of traditional data?

The objective of the workpackage WP1: “Representation and retrieval of non-traditional data” is to provide representational models for non-traditional data that are metadata-rich, along with mechanisms for the automated extraction of this meta-information for new data, and indexing techniques that will facilitate the efficient retrieval of data on the basis of this meta-information.

• Task T1.1 introduces extensible representation models for non-traditional data (specifically, semi-structured, streaming and spatiotemporal data), with an emphasis to their meta-information (in terms of semantics, content and structure).
• Task T1.2 introduces mechanisms for the automatic extraction of meta-information (concerning content, structure, origin etc), which is typically hidden within the data, for the three aforementioned types of data.
• Task T1.3 explores the integration of existing indexing mechanisms into the dataspace management system, as well as the introduction of novel indexing mechanisms, wherever appropriate (since the coexistence of all these data within the same dataspace allows their joint management).

How do we allow transparent information synthesis?

The objective of the workpackage WP2: “Data integration, navigation and retrieval” is to provide the theoretical foundation, the algorithmic techniques and the software architecture for the definition/composition of dataspaces and the usage of both traditional and non-traditional data in their context.

• Task T2.1 concerns the definition of a central, generic metamodel for data in a dataspace, along with appropriate abstractions that allow different types of (both traditional and non-traditional) data to be mapped to this generic metamodel. The overall effort will be based on model management and model mapping techniques that will allow the management of data and the tracing of their interdependencies.
• Task T2.2 facilitates the automated extraction of mappings between the data in a dataspace via the introduction of mapping discovery and mapping composition algorithms that exploit the aforementioned metamodel and the meta-information of the integrated data.
• Task T2.3 introduces the software architecture and algorithms for (a) unified access and retrieval for the data of a dataspace (to be performed in a generic fashion that hides the particularities of data format and the query language that is appropriate for each kind of data) and (b) the gradual training of the system to automate as much as possible the addition of new data sources.

How do we attack the impersonal, pret-a-porter querying?

The objective of the workpackage WP3 “Contextualization and personalization of information” is to define models and algorithmic techniques for the support of personalized and contextualized services in a dataspace.

• Task T3.1 concerns the extraction of contextual and semantic information for the user workspace, in a generic model that captures all user actions (as recorded in the system’s log) along with user patterns that are mined from the above tracing and profile management operators.
• Task T3.2 introduces a graph-based, generic model of user preferences and profiles, along with techniques for the interpretation of user patterns on the basis of user preferences. Moreover, the task provides techniques for the combination of different profiles for the same user as well as techniques that allow the sharing of the same user profile to different applications of the ecosystem.
• Task T3.3 explores the design of a personalization system over the previously attained results; the system is equipped with operations for profile management, user recommendation, query result personalization and diversification.

How do we handle scale via distribution?

The objective of the workpackage WP4: “Distributed Infrastructure for the storage, retrieval, and management of data” is to provide architectures, algorithms and facilitating data structures for the distributed organization of dataspaces at the lower levels of a dataspace management engine.

• Task T4.1 is centered on the efficient profiling of users and the support of their querying tasks by distributing computation and query processing over a scalable architecture. The task will introduce methods for exploiting collaborative wisdom in order to construct taxonomies of the collected data along with user profiles. Also, the task will provide query processing algorithms that exploit the above in order to efficiently answer the typically encountered ranking and join queries over a distributed infrastructure.
• Task T4.2 involves the construction and exploitation of statistical information -in the form of data synopses- for different kinds of data (with particular emphasis to uncertain streaming data) and its incorporation in query processing algorithms for cloud infrastructures.
• Task T4.3 proposes algorithms that compute the similarity of users of the ecosystem, in order to provide recommendations on the basis of the previously attained results. These similarity results are also extended to cover the similarity of nodes of the ecosystem too, to facilitate more efficient query processing. Finally, the task will provide algorithms for the efficient maintenance of “neighborhoods” of similar users and data over time.

How do we design for and adapt to change?

The objective of the workpackage WP5: “Evolution and self-tuning of information ecosystems” is to provide algorithms and mechanisms for the design and adaptation of a dataspace and its ecosystem to changes that concern the semantics, structure and content of the stored information.

• Task T5.1 provides models and languages for the management of evolution of dataspaces and their ecosystems, with particular emphasis to the modeling of the internal structure and content of the ecosystem (over which change will occur), the treatment of change as “first class citizen” (in order to facilitate provenance management) and the query language via which the administrators and data curators can navigate in different variants of the dataspace contents.
• Task 5.2 provides algorithms for the adaptation and self-tuning of the dataspace in the presence of change. Emphasis is given to the efficient and correct restructuring of the mappings between objects of the dataspace once change occurs either to the schema or to the content of an object in the dataspace.
• Task 5.3 deals with the problem of designing an ecosystem with a view to evolution and provides (a) design metrics that allow the assessment of the quality (i.e., adaptability) of the design of an existing ecosystem and (b) design algorithms that derive the best possible design having specific evolution scenarios in mind.

The project coordinator is responsible for coordinating the work of collaborating research groups as well as to monitor the technical scope and budget to ensure the smooth running of the project. As part of this action will.

• provide scientific guidance,
• control the quality of deliverables,
• ensure that there are appropriate resources to complete tasks within prescribed timeframes,
• helps solve any problems,
• facilitate collaboration between research groups,
• ensures the continuity and interoperability between the results of work packages.

The project's evaluation will be held upon the completion of the project. Reviewers will be distinguished members of the data management research community and will be invited by the Coordinator and the Steering Committee (composed of the coordinators of the research teams 5) the project. The package deliverable will be the final evaluation report at the end of the project, which will include reviews on the project results and recommendations for their further exploitation.