The Role of Space and Time For Knowledge Organization on the Semantic Web

Submitted by Pascal Hitzler on 05/06/2010 - 08:39
Paper Title: 
The Role of Space and Time For Knowledge Organization on the Semantic Web
Authors: 
Krzysztof Janowicz
Abstract: 
Space and time have not received much attention on the SemanticWeb so far. While their importance has been recognized recently, existing work reduces them to simple latitude-longitude pairs and time stamps. In contrast, we argue that space and time are fundamental ordering relations for knowledge organization and reasoning. While most research on Semantic Web reasoning has focused on thematic aspects, this paper argues for an unified view combining a spatial, temporal, and thematic component. Besides their impact on the representation of and reasoning about individuals and classes, we outline the role of space and time for ontology modularization, evolution, and the handling of vague and contradictory knowledge. Instead of proposing yet another specific methodology, the presented work illustrate the relevance of space and time using various examples from the geo-sciences.
Full PDF Version: 
PDF icon swj34.pdf
Submission type: 
Other
Responsible editor: 
Pascal Hitzler
Decision/Status: 
Accept
Reviews: 

Review by Christoph Schlieder:

This well-written position statement argues that the spatial and temporal context of conceptualizations induces semantic heterogeneity which should be considered a feature rather than a bug. A number of interesting issues are raised:

(1) The author calls for a shift in perspective from resolving semantic heterogeneities towards acknowledging the importance of local conceptualizations.

Convincing examples are provided showing how and why local conceptualizations emerge. The different conceptualizations of potholes in Britain are especially intriguing. I wonder, however, whether there is not a common core to these concepts that might be formalized: "A pothole is a crack in a road that the local council is bound to repair because of its size."

It seems that there is a need for both, local and global semantics, or, in terms of the examples given, for potholes as for anemometers. The attitudes of cultivating heterogeneities and eliminating them are rather opposed, though. Do you see any way that the two attitudes could coexist?

(2) The context of ontological statements should no longer be neglected: the Web's triple A becomes a quintuple, AAA at Any time and Anywhere.

In the study of natural languages, syntax and semantics are supplemented by pragmatics which studies in what way the context of an utterance contributes to its meaning. Is the literature on pragmatics relevant to the case made in the article?

(3) It is argued that embodiment provides a key to the problem of modeling context. The author advances the idea of grounding the Semantic Web in the Sensor Web.

While perceptual grounding is certainly relevant and supported by research results from cognitive semantics, there is also a social aspect of semantics. Meaning can be the matter of negotiations between members of a community in which they seek for the clarification of ambiguities. Is there a place for negotiated semantics in your picture of the future Semantic Web?

Typos
p.5 believes -> beliefs

Tags: 
Reviewed

Comments

Submitted by Pascal Hitzler on 06/12/2010 - 18:50.

I read the position statement with pleasure. In particular the clear examples from the geoinformatics domain are very enlightening. They are indeed excellent evidence for the main point in the paper, namely that suitable use of context information is crucial in some areas. While this is by no means a new insight, the paper makes an excellent case that the geoinformatics domain, due to an appropriate mixture of formal/crisp and informal/vague/fluent notions, would be a very suitable area to start develop feasible ways of modeling context.

The paper also makes an excellent case that space and time play a central role in modeling context, in particular in geosciences. I was wondering though, to what extent this centrality carries over to other domains.

In contrast to this excellent motivation, the second part of the paper remains a bit vague. While the microtheories approach, which is mainly discussed there, appears to be one viable approach to the context modeling problem (and is strongly related to McCarthy's "ist" predicate), I would like to see a bit more evidence, or arguments, or at least pointers, why the author suggests this approach (and not others) as a promising line to follow.

Minor remarks:

abstract:

an unified -> a unified

illustrate -> illustrates

p2 right: abandonment -> abandoned

p3 left:

for navigation purpose*s*

concerning the last sentence of the paragraph in that column ("Hence, on-the-fly ..."): I would like to know a bit more about the author's arguments why these are central.

p3 right: specification -> specifying

p4 left: last sentence of the paragraph ("Personal information managers ...") seems to be broken

p4 right:

C-OWL[21] -> C-OWL [21]

an promising -> a promising

is not randomly -> is not random

which kinds of -> which kind of

Sentence "Separate microtheories may..." seems to be broken

p5 left:

line after top equation should not be indented

I'm puzzled about the second equation and the subsequent sentence. A bit more explanation about genlMtC would be helpful.

demonstrate*d*

p5 right:

natural kinds evolve (what is meant here?)

creation of *an* isolated island