Allowing contributors to enter alternative viewpoints with provenance
Provenance is key for establishing trust. Trusting information implies a judgment by the consumer on whether they accept the information. Trust is based on provenance records, for example looking for authoritative sources in the provenance records or particular well-accepted processes such as a published scientific method. In social knowledge collection frameworks, the provenance of the information really matters since the contributions can have varying quality.
Traditional research on trust focuses on entity trust, where the trust is associated with particular entities (Artz and Gil 07). For this work, we are interest in a novel notion of content trust, where the trust is placed on a piece of information (Gil and Artz 07). A qualitative notion of content trust can be provided as users create queries that refer to provenance records of entities. For example, “According to government sources, what US congress representatives own a business?” would deliver to the user only answers that were associated with sources that are government organizations (e.g., NIST). We can also measure trust quantitatively with metrics derived from provenance records.
For this work, we are looking at tracking provenance in sites that crowdsource geospatial information, and in mashups of geospatial sources. We are working with the Open Geospatial Consortium to understand requirements for provenance in a geospatial context, and to determine how quality and trust can be derived from provenance records. We are using the W3C PROV standard to represent geospatial provenance. An OGC Provenance Report will be released publicly soon, for now there is an interim OGC document available to OGC members: the OGC OWS-10 Provenance wiki.
This work is reported in the following publications:
* “Challenges in Modeling Geospatial Provenance." Daniel Garijo and Yolanda Gil and Andreas Harth. In Proceedings of the Fifth International Provenance and Annotation Workshop (IPAW), Cologne, Germany, 2014. Available as a preprint.
* "User Requirements for Geospatial Provenance." Daniel Garijo, Yolanda Gil, and Andreas Harth, A. Provenance Analytics, co-located with the Fifth International Provenance and Annotation Workshop (IPAW), 2014. Available as a preprint.
* "Geospatial Data Integration with Linked Data and Provenance Tracking." Andreas Harth and Yolanda Gil. W3C/OGC Workshop on Linking Geospatial Data, 2014. Available as a preprint.