english

If you experience slow out-of-cache responses with the stock BIND (9.2.3) nameserver on SuSE Linux Enterprise Server (SLES9), try to disable to ipv6 kernel module. This seems to be related to a similar issue in CentOS

Is it just me or does the term "language-independent" (Named Entity Recognition, oder whatever) seem a bit pompus? When in fact all papers I've seen on the subject instead suggest ways to train multiple single-language classifiers - provided that you've got corpora for all languages and even genres.

A while ago I started managing my bookmarks (those that are suitable for the public, mind you) with del.icio.us. I really like it, and I'd like to also put it to use for the bibliography for my masters thesis, but del.icio.us has a big drawback for that: It doesn't seem to support any electronic bibliography formats like e.g. BibTeX.
Furthermore, scientific publications have some more metadata (like the author) one would like to have searchable, and not in the regular tags please (if you think it should be in the tags, please tell me why.)

There are indeed alternatives focusing on scientific publications, and - as with all great ideas and their good follow-ons - even more than one. Even though choise is great in principle, I'm not too happy about this as for social software, you want to have network effects and the spread of users over several sites hurts for that matter.
And I have to decide: Do I stay with del.icio.us and it's large user base, even though it has no support for citation software? (maybe abuse the the description field for BibTex data). Or switch to a more scientifically oriented service? And which to chosse? I know of Connotea, BibSonomy and CiteULike. Any recommendations?

Ian Holsman asks : "should OSS care about the late majority/laggards"?
At least the PHP project seems to answer "NO", and I have to admit I'm not too happy about that.

In my opinion, the question whether an OSS project should care about the late majority is a piece of whether the project will be seen as "ready for the enterprise", as large enterprises tend to the late adopter side.
However, the issue is not only about late adoption, it is also very much about systems that stay in service for quite a long time and where large changes to the infrastructure are avoided whereever possible. This is also a characteristic of "enterprise" deployments.

Thus, not offering long-term support as an OSS project could influence enterprises to opt for other software infrastructure. Furthermore, I believe the argument that independent consultancies could provide long-term support even when the main project has abandoned the software version in question is somewhat moot, as many will have the feeling that independent consultants will not be able to offer the same quality as the original project.

"The matter of knowing when to stop in hierarchical agglomerativ clustering has been explored, but not resolved."[1]

That sums up my current problems pretty well.

[1] D. Beeferman , A. Berger: "Agglomerative clustering of a search engine query log". Proc. of the 6th ACM SIGKDD conference, 2000