 |
| Titel |
An algebra for spatio-temporal information generation |
| VerfasserIn |
Edzer Pebesma, Simon Scheider, Benedikt Gräler, Christoph Stasch, Matthias Hinz |
| Konferenz |
EGU General Assembly 2016
|
| Medientyp |
Artikel
|
| Sprache |
en
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 18 (2016) |
| Datensatznummer |
250129414
|
| Publikation (Nr.) |
 EGU/EGU2016-9523.pdf |
|
|
|
|
|
| Zusammenfassung |
| When we accept the premises of James Frew’s laws of metadata (Frew’s first law: scientists
don’t write metadata; Frew’s second law: any scientist can be forced to write bad metadata),
but also assume that scientists try to maximise the impact of their research findings, can
we develop our information infrastructures such that useful metadata is generated
automatically? Currently, sharing of data and software to completely reproduce
research findings is becoming standard, e.g. in the Journal of Statistical Software [1].
The reproduction (e.g. R) scripts however convey correct syntax, but still limited
semantics.
We propose [2] a new, platform-neutral way to algebraically describe how data is
generated, e.g. by observation, and how data is derived, e.g. by processing observations. It
starts with forming functions composed of four reference system types (space, time, quality,
entity), which express for instance continuity of objects over time, and continuity of fields
over space and time. Data, which is discrete by definition, is generated by evaluating such
functions at discrete space and time instances, or by evaluating a convolution (aggregation)
over them. Derived data is obtained by inputting data to data derivation functions, which for
instance interpolate, estimate, aggregate, or convert fields into objects and vice
versa.
As opposed to the traditional when, where and what semantics of data sets, our algebra
focuses on describing how a data set was generated. We argue that it can be used to discover
data sets that were derived from a particular source x, or derived by a particular procedure y.
It may also form the basis for inferring meaningfulness of derivation procedures [3]. Current
research focuses on automatically generating provenance documentation from R
scripts.
[1] http://www.jstatsoft.org/ (open access)
[2] http://www.meaningfulspatialstatistics.org has the full paper (in review)
[3] Stasch, C., S. Scheider, E. Pebesma, W. Kuhn, 2014. Meaningful Spatial
Prediction and Aggregation. Environmental Modelling & Software, 51, 149–165 (open
access) |
|
|
|
|
|
|