Secondo Plugins: A Platform to Publish Research Implementations
===============================================================
A significant part of research in databases concerns the development of new
data structures or algorithms, e.g. for indexing or query processing. Whereas
the papers describing these structures are easily accessible after publication,
the software developed and used in the experiments is all too often lost.
This is unfortunate for several reasons. Authors of later papers who try to
offer an improved solution need to reimplement the previous software - a
process that is not only a waste of effort but also error-prone. What is
compared to may not be the exact proposal by the original authors. Second,
readers of the paper cannot repeat the experiments or do other experiments.
Third, the software cannot be used in a system context.
We offer a platform to publish such software together with the paper
describing it, as a so-called Secondo Plugin. Secondo is an extensible DBMS
prototype developed for many years at University of Hagen. It is extensible at
all levels including kernel system, query optimizer and graphical user
interface. In particular, the kernel is extensible by so-called algebra
modules which offer collections of data types and operations. Secondo
provides fairly sophisticated management of spatial data and moving objects.
An index structure or query processing algorithm developed in research may be
offered within a new algebra module as a type or operator, respectively.
Operators can be called directly at a query interface below the optimizer.
This allows one to easily test indexes or algorithms without the need to
circumvent the query optimizer (i.e. before integration into query
optimization has been achieved).
The recently added feature of Secondo plugins enables anyone to publish a
Secondo extension, independently from the Secondo team. Essentially,
extension components need to be wrapped into a zip file. This includes an XML
file describing which extensions have to be put where within a standard
Secondo distribution.
The plugin may then be published together with the paper, e.g. on the
authors’ web site. A reader of the paper may get a Secondo system from
University of Hagen and the plugin from the authors’ web site. A small
installation program integrates the plugin into the system.
We invite you to publish your new research implementations in this way.
Advantages are the following.
- Readers of your paper will be able to repeat your experiments, especially
if you provide Secondo scripts (files with Secondo commands) executing
related queries. They may also do other experiments that you did not think
of, or use other data sets.
- Authors of later papers will compare to your algorithm rather than others
(when having a choice), because this is easy to do. They will compare to
your correct implementation, not to what they misunderstood you
implemented.
- This in turn will lead to more citations of your paper.
- Your software will go into a pool of query processing software and hence
become more popular than otherwise. It may even be used for practical
applications.
See the Secondo system at
http://dna.fernuni-hagen.de/Secondo.html/index.html.
The Secondo Plugin concept is explained via that site or directly at
http://dna.fernuni-hagen.de/Secondo.html/start_content_plugins.html
At the site some plugins are shown that are available already. When you
publish a plugin, you are welcome to notify us so that we can include a
link on our web site.
Ralf Hartmut Güting and the Secondo team
University of Hagen
http://dna.fernuni-hagen.de/gueting/home.html