from domain experts and in terms of the computational complexities that arise when the sources are not trivially compatible. Consequently, most existing solutions rely on a one-size-fits-all approach, where the data are integrated once (i.e., the cost of the integration is paid upfront) and then the integrated data or knowledge-bases are used as is, until a new data source is made available. However, such snapshotbased integration solutions cannot be effectively applied when the data sources are autonomous and dynamic; instead data should be integrated on an as-available and as-needed basis. Furthermore, users of the integrated data --e.g., scientists, decision makers-- often have a high degree of knowledge of the domain and strong beliefs about the kinds of integration operations that would be acceptable. They are thus indispensable to the integration process: their needs, assumptions, and knowledge must be fully leveraged. In particular, we observe that overly-eager, early conflict resolution (where some alternative interpretations are deemed inapplicable without sufficient evidence) may be detrimental to the effective exploration and use of the available knowledge.
|Effective start/end date||8/15/10 → 7/31/15|
- National Science Foundation (NSF): $499,946.00