Anzeige
Erschienen in:
2017 | OriginalPaper | Buchkapitel
A Framework for User-Driven Mapping Discovery in Rich Spaces of Heterogeneous Data
verfasst von : Federica Mandreoli
Erschienen in: On the Move to Meaningful Internet Systems. OTM 2017 Conferences
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
Data analysis in rich spaces of heterogeneous data sources is an increasingly common activity. Examples include exploratory data analysis and personal information management. Mapping specification is one of the key issues in this data management setting that answer to the need of a unified search over the full spectrum of relevant knowledge. Indeed, while users in data analytics are engaged in an open-ended interaction between data discovery and data orchestration, most of the solutions for mapping specification available so far are intended for expert users.
This paper proposes a general framework for a novel paradigm for user-driven mapping discovery where mapping specification is interactively driven by the information seeking activities of users and the exclusive role of mappings is to contribute to users satisfaction. The underlying key idea is that data semantics is in the eye of the consumers. Thus, we start from user queries which we try to satisfy in the dataspace. In this process of satisfaction, we often need to discover new mappings, to expose the user to the data thereby discovered for their feedback, and possibly continued towards user satisfaction.
The framework is made up of (a) a theoretical foundation where we formally introduce the notion of candidate mapping sets for a user query, and (b) an interactive and incremental algorithm that, given a user query, finds a candidate mapping set that satisfies the user. The algorithm incrementally builds the candidate mapping set by searching in the dataspace data samples and deriving mapping lattices that are explored to deliver mappings for user feedback. With the aim of fitting the user information need in a limited number of interactions, the algorithm provides for a multi-criteria selection strategy for candidate mapping sets. Finally, a proof of the correctness of the algorithm is provided in the paper.