Impoverished Identification by URI
There is one final part of the faliure of the Semantic Web puzzle to explore before we can talk about a solution.
In owl:sameAs and Linked Data: An Empircal Study, Ding, Shinavier, Finin and McGuinness write:
Our experimental results have led us to identify several issues involving the
owl:sameAsproperty as it is used in practice in a linked data context. These include how best to manageowl:sameAsassertions from “third parties”, problems in merging assertions from sources with different contexts, and the need to explore an operational semantics distinct from the strict logical meaning provided by OWL.
To resolve varying usages of owl:sameAs, the authors go beyond identifications provided by a URI to look to other properties. For example:
Many owl:sameAs statements are asserted due to the equivalence of the primary feature of resource description, e.g. the URIs of FOAF profiles of a person may be linked just because they refer to the same person even if the URIs refer the person at different ages. The odd mashup on job-title in previous section is a good example for why the URIs in different FOAF profiles are not fully equivalent. Therefore, the empirical usage of owl:sameAs only captures the equivalence semantics on the projection of the URI on social entity dimension (removing the time and space dimensions). In thisway, owl:sameAs is used to indicate p artial equivalence between two different URIs, which should not be considered as full equivalence.
Knowing the dimensions covered by a URI and the dimensions covered by a property, it is possible to conduct better data integration using owl:sameAs. For example, since we know a URI of a person provides a temporal-spatial identity, descriptions using time-sensitive properties, e.g. age, height and workplace, should not be aggregated, while time-insensitive properties, such as eye color and social security number, may be aggregated in most cases.
When an identification is insufficient based on a single URI, additional properties can be considered.
My question then is why do ordinary users have to wait for experts to decide their identifications are insufficient? Why can’t we empower users to declare multiple properties, including URIs, as a means of identification?
It could be something as simple as JSON key/value pairs with a notation of “+” for must match, “-” for must not match, and “?” for optional to match.
A declaration of identity by users about the subjects in their documents. Who better to ask?
Not to mention that the more information supplies with for an identification, the more likely they are to communicate, successfully, with other users.
URIs may be Tim Berners-Lee’s nails, but they are insufficient to support the scaffolding required for robust communication.
- The Semantic Web Is Failing — But Why? (Part 1)
- The Semantic Web Is Failing — But Why? (Part 2)
- The Semantic Web Is Failing — But Why? (Part 3)
- The Semantic Web Is Failing — But Why? (Part 4)
- The Semantic Web Is Failing — But Why? (Part 5) This post.
The next series starts with Saving the “Semantic” Web (Part 1)
[…] from the adventures of owl:sameAs (The Semantic Web Is Failing — But Why? (Part 5)) that any single string identifier is subject to multiple interpretations. Interpretations that can […]
Pingback by A Tale of Five Languages « Another Word For It — February 11, 2013 @ 10:59 am
[…] The Semantic Web Is Failing — But Why? (Part 5) […]
Pingback by Simple Web Semantics – Index Post « Another Word For It — February 18, 2013 @ 4:23 pm