Pinned toot

IDF.social is your home if you know:

\[
\textit{idf} = - \log \frac{n_t}{N}
\]

and... everyone else!

Pinned toot

Dear all,

For now, I will keep the site closed for registrations; but everyone can independently invite people to join using the following link to generate an invite:

idf.social/invites

If you know someone who wants to join this instance, then please do give them an invite!

Consider the instance open for those who know anyone on IDF.social.

"Sense of community" and an initial seed of some 40 IR researchers should be enough to keep those spammers out of here and still grow the club!

Pinned toot

Computer scientist 👨‍🏫 with main focus on Information Retrieval 📄

Also interested in Data Science / Big Data 💾

In the previous millennium :mastodon: a PhD on Multimedia Databases 🎥 (Yes that was a respectable research topic once upon a time pre-birdsite!)

is my academic home.

Entrepreneur: co-founder of (spinque.com)

IDF.social admin.

Loves indie music: follow me at @arjenpdevries if you are interested in gigs, events, etc.

Pinned toot
Pinned toot

Inaugural toots on idf.social using a photo of lunch with Nick Belkin after attending the now Dr. Hugo Huurdeman defense.

See also:
timelessfuture.com/phd

[CfP] SustaiNLP: First Workshop on Simple and Efficient Natural Language Processing sites.google.com/view/sustainl

Remember when a Google #search used to lead you somewhere?

Now it increasingly just keeps you on Google. In fact, Google results take up 62.6% of the first screen of search results in a sample of 15,000 searches.

#Google is building a walled garden that encompasses some of the most accessed information on the internet with no recourse for businesses who are suddenly finding themselves on the wrong side, cut off from their users.

themarkup.org/google-the-giant

#privacyMatters #searchengine

MARGE: Pre-training via paraphrasing by keynote speaker Luke Zettlemoyer.

arxiv.org/abs/2006.15020

5. Ignore results for collections when there are no baseline from independent research;
6. Test collections wear out! Expected maximum result increases with number of runs (on see leaderboards! -- Carterette's SIGIR paper)
7. Conferences and journals need to accept papers with "null" results. (to prevent the busy beaver / the p-hackers) Reproducibility is important: Publish your code and data
8. Evaluation initiatives are important (but they should only run proper measures and methods)

Norbert Fuhr's recommendations for gaining scientific knowledge from experiments:

1. Do not use MRR or MAP;
2. Instead of relative improvements, regard the effect size!
3. For multiple significance tests, use a correction, such as Bonferoni or Tukey's HSD (NB comparing only to the 2nd best method does not help!)
4. There are no significant improvements for re-usable test collections! (hypotheses have to be formulated before the work)

In retrospect it's weird that EdgeHTML turned out to be this brief transitional engine that only existed between 2015-2019. Even Opera Mini and UC Browser seem to have lasted longer. KaiOS may last longer still.

@nfuhr or, is this a wannabe - I'd understand, I wannabe Norbert too 😉

Exciting new accountname in the fedishpere: @nfuhr

Will it be the great Salton Award winner, this wise man from Duisburg-Essen?

Show more
Mastodon

The "unofficial" Information Retrieval Mastodon Instance.

Goal: Make idf.social a viable and valuable social space for anyone working in Information Retrieval and related scientific research.

Everyone welcome but expect some level of geekiness on the instance and federated timelines.