Luke
Lucene Index Browser.
Launch with Java WebStart
License: Apache 2.0 |
Lucene is an Open Source,
mature and high-performance Java search engine. It is highly flexible,
and scalable from hundreds to millions of documents (the tests I know
about used 43 mln documents, achieving < 10s search times).
Luke is a handy development and diagnostic
tool, which accesses already existing Lucene indexes and allows you to
display their contents in several ways:
- browse by document number, or by term
- view documents / copy to clipboard
- retrieve a ranked list of most frequent terms
- execute a search, and browse the results
- selectively delete documents from the index
- and more...
|
Stempel
An algorithmic stemmer and
lemmatizer for the Polish language.
License: Apache 2.0 |
Stemming and lemmatization are
important tools for Information Retrieval. They help to improve recall
, especially in case of highly-inflectional languages like the Slavic
family.
Unfortunately, in case of Polish language there is only one freely
available dictionary-based stemmer by Dawid
Weiss, and no open source algorithmic stemmers exist. This project
aims to fill this gap with a high quality open source implementation,
based on algorithmic approach. The algorithm and its implementation
come from the Egothor project.
This project also provides stemming tables built from high-quality
corpora.
Current version of the stemmer achieves ca. 95% accuracy of stemming,
for previously unseen word forms, and ca. 75% accuracy of
lemmatization. This means that even at this early stage it is very
useful. |
Murmur Hash
License: Apache 2.0
|
A Java implementation of a fast
hash function, created by Austin Appleby (see
this site
for more information and discussion of its performance).
This is a very fast hash, with excellent avalanche behavior. Compared with
the FNV1a32 (see below) it's roughly 10 times faster, and roughly
5 times faster that a Java version of Jenkins' hash (
available here).
This implementation is a Java port of a C version of MurmurHash 2.0 .
|
FNV1 Hash
License: Apache 2.0
|
A Java implementation of fast
hash functions, originally created by Glenn Fowler, Phong Vo, and
improved by Landon Curt Noll.
"FNV1 hashes are designed to be fast while maintaining a low collision
rate. The FNV1 speed allows one to quickly hash lots of data while
maintaining a reasonable collision rate. The high dispersion of the
FNV1 hashes makes them well suited for hashing nearly identical strings
such as URLs, hostnames, filenames, text, IP addresses, etc."
This is a straightforward port of the public domain C version, written
by Landon Curt Noll (one of the authors), available from his website.
UPDATE: please see
this discussion
on the poor avalanche behavior of this hash. A better alternative would be
the Jenkins' hash or the MurmurHash. |
Thinlet
Lightweight Java GUI framework,
based on XUL concept.
License: Lesser GPL |
Thinlet is a small one-class XUL
engine, which allows you to build fairly complex Java GUI applications,
and to keep the "View" part in MVC (Model-View-Controller)
well-separated from your business logic. Thinlet-based applications are
characterized by small memory footprint for GUI (Thinlet engine weights
less than 50kB), pleasant look'n'feel, and rapid development cycle.
Please see my project Luke for a non-trivial demo
of an application based on Thinlet, or visit Thinlet home page to see
other demos.
I also successfully developed a few mobile applications based on a J2ME
version of Thinlet. |
ebXML / ebTWG
UN/CEFACT effort to build a
framework enabling enterprises of any kind to conduct business
electronically. |
Ongoing work (will it ever
end...) on standardizing business models and interfaces for e-commerce.
I was involved in the discussions on the use of Resource-Event-Agent
(REA) economic modeling framework in ebTWG models. REA framework
provides some of the central concepts there... You can take a look at REA Ontology page where you can find
a formal ontology in Protege
format. You may want to visit also my ebXML Ontology page. |
ECIMF
(E-Commerce Integration
Meta-Framework)
EU standardization project in
the area of e-commerce interoperability |
This 18-month project attempted
to come up with high-level meta-modeling framework and set of
guidelines for successful e-commerce integration between different
standards. The mirror of the original project website can be found here, although the final deliverables can only be
obtained from the CEN / ISSS website. |
PicoBSD
One-floppy version of FreeBSD, for embedded applications
License: BSD
|
I started this project mostly
just to see if it's possible to squeeze FreeBSD into 1.44MB - prompted
by the famous QNX demo floppy. To make the long story short, it was
possible to do it :-) although it involved writing from scratch many of
the basic system utilities (init, shell, ps, vm, netstat etc). I even
created a version of the floppy with a simple windowing system (called
'W'), which included a graphical WWW browser... It wasn't so
straightforward at the time, and I learned a couple of tricks on how to
minimze the space demands of FreeBSD while keeping the functionality...
Later on I realized that such minimalistic version would work quite
well also in embedded environments, and so the project went on this way
from now on...
I'm no longer actively involved in the project - the official website
is therefore horrendously out of date. However, the project is still
active, and the best source for most up-to-date PicoBSD is the official
FreeBSD source tree, under src/release/picobsd. There is
also a dedicated mailing list freebsd-small@freebsd.org
where you can find some help.
|
FreeBSD
A powerful Unix-derived
operating system.
License: BSD
|
Apart from being involved in
PicoBSD project, I also made some contributions to the other parts of
FreeBSD. I was one of the strong supporters of using Forth for the new
bootloader design, I also implemented a subset of standard "cons25"
terminal emulation in the bootloader. Other significant contribution
was to implement ability to create "sysctl" nodes dynamically (before
they were created statically using linker sets, during kernel linking).
FreeBSD is a very powerful platform for server applications, and I'm
using it frequently in situations where a Unix-like server is required.
I like the clean and well thought-out layout of the system, consistent
design. If I should mention one fundamental aspect of FreeBSD that sets
it apart from Linux (besides the business-friendly license, of course
:-) then it would be the fact that FreeBSD is being developed as a
complete system, i.e. not only kernel but also with all standard Unix
tools, and that all sources to all parts of the system are distributed
as a complete, tested and ready to build tree - which makes the
maintenance of FreeBSD systems a breeze. If you add to this the famous
"ports" system - only now being copied by other OS-es - then I think
you have a clear winner. But I'm biased, of course... :-)
|
