Talk:Semantic MediaWiki/Open issues

Annotating time

I was quite impressed with how Wikipedia models time already. The Year/Decade/Century pages look to me like the beginning of a semantic annotation already, so I think it would be a pity to ignore them. As you say - this could very easily become over complicated - but I would suggest using the exiting year templates to limit what would be done. Rattle 23:21, 21 August 2005 (UTC)

I partially agree. Some of the time-table information in Wikipedia can already be represented in our current approach: one can have a data type for dates and associate them with properties such as "birthdate". This enables us both to create time tables and to reuse them for semi-automatic annotation. The unsolved problem is that one can only annotate articles (via properties) with some time, but one cannot directly describe that an annotation belongs to some fixed time only. This kind of data is also in Wikipedia, but it is hard to describe it in our framework (as in most common frameworks for knowledge representation!). The reason is that annotations cannot again have properties, and that, even if they could, it would be cumbersome to annotate annotations (even if just with simple time values). So to keep the Wikisource simple, we currently do not include timing of annotations into the project, but we can have time-related properties for articles. --Markus Krötzsch 18:30, 25 August 2005 (UTC)

Maybe I got the point wrong, but what do you mean by "fixed time"? There is a discussion concerning properties in categories, which seems to overlap with the semantic approach. There a solution for the problem of having lists for the highest mountains in New Zealand was the starting point. It ended with the suggestion of storing data into properties defined by categories that contain the article name as the primary property. This approach helps to limit the properties and make them consistent within a group of articles.

What I am also confused with is the definition of uncertainties, ranges or statistical values. Is'nt it possible to define such ranges. In that example with the population of the city the size should be also prametrised in time, since population numbers are changing continuously. Thereby it would make sense to define a mean value and an uncertainty value, which would resolve the problem - whenever a list is generated it will be the mean that counts, but the paired uncertainty would have to be displayed as well. So is it possible to build more complex data types out of simple ones?

Another thing yet is the implementation of quasi-static searches, with templates. This allows for a dynamic table in an article like: "20 highest montains in New Zealand in metres".

On the other hand it would also be nice to access the predecessor and successor in a list (property of a category) which allows for generation of dynamic navigation elements or windowed time bars. --BoP 23:32, 29 September 2005 (UTC)

Former capitals

"For the beginning, it might be helpful to include (very coarse) temporal information in link types and data-fields. E.g. one can have a relation "was capital of" or a data-field "former capital" in addition to the current ones."

Could we in a systematic way build a relation former-r from relation r, by defining former-r as (r and former), if we have a former or in the past predicate or category ? Apokrif 14:25, 11 August 2006 (UTC)