Microsoft Research Publications | Hacker News Reader

Microsoft Research Publications

research.microsoft.com

186 pointsLegend28 comments

Open articleView on HN

How is it that they can say "Our research" about publications from 1951, like:

http://research.microsoft.com/apps/pubs/default.aspx?id=6500...

and

http://research.microsoft.com/apps/pubs/default.aspx?id=6503...

are they just part of Microsoft's IP portfolio, or are these authors previous incarnations of Bill Gates?

Nice find. Sorting by oldest first turns up pages and pages of classic papers, e.g. by Kolmogorov, Bucy, David Marr, etc., etc. Many of these people have never had a relationship with MS. And they are, in many cases, copyright by other entities, like the IEEE or the Royal Statistical Society, so they are in no way MS IP.

This is just a guess, but maybe those are papers that are cited in others from MS Research authors, and there was a data migration bug at some point that mixed citations with authored papers.

It is obviously a bug since MSR was founded in...1991. Many researchers also insert papers into the system written in other labs or universities before joining MSR; these seem to be included as well even though these days we clearly get to distinguish between was written when at MSR and what wasn't. This behavior seems to only apply to older papers and not newer ones.

I didn't know they made videos of visiting speakers available:

http://research.microsoft.com/apps/catalog/default.aspx?p=1&...

There's also an RSS feed:

http://research.microsoft.com/rss/visitingSpeakers.xml

I'm curious about this paper: The use of Melodic Scales in Bollywood Music: An Empirical Study

Cool that Microsoft does research into Bollywood music?

http://research.microsoft.com/apps/pubs/default.aspx?id=1983...

http://research.microsoft.com/en-us/um/redmond/projects/song...

Could it be related to SongSmith?

"Songsmith generates musical accompaniment to match a singer’s voice. Just choose a musical style, sing into your PC’s microphone, and Songsmith will create backing music for you."

I took an interdisciplinary course in University and one of the CS professors did a whole lecture series on research in music and analysis. A lot of beat analysis, understanding rhythms and predicting popularity of songs based on the distribution of the timing between their beats. It was very cool and extremely creative.

Here's his homepage if you want to learn more http://cgm.cs.mcgill.ca/~godfried/

Edit: Looks like he recently published a book on it too http://www.crcpress.com/product/isbn/9781466512023

His research is definitely fascinating. Funny to see my CS professor pop up in here though. :)

Be aware - some algorithms described in publications on research.microsoft.com can be patented by microsoft (and this fact not referred in publication itself, google patent search can be useful for this).

How would it go if you were to write an algorithm based on the patented version?

Very secretly. On a serious note, I wonder what would it mean to implement a patented algorithm (patented in US where it's possible) in a country where it's not possible to patent software, like anywhere in EU. I presume it would mean you couldn't sell your software in US and that's about it?

How interesting.

Isn't there a requirement to mark every publication of a patented idea? Did Microsoft just invalidated those patents? (Do they even know what patents apply?)

I'm not an academic, but I read a lot of scholarly publications in computer science and I have never seen a single reference to a patent in any of them. Usually I only discover that something is patented or patent pending later, after some heavy Googling.

I copied down this quote from somewhere earlier this year but I can't find the source now: "One of the things they tell us developers here at Microsoft is that we should never read any patents. That's because you're liable for triple damages if you knowingly infringe on a patent."

Yes, I went through that training, and that's an accurate summation of the commandments given. Do not peruse patent databases because it only takes one person at Microsoft reading the patent, and someone else totally unrelated infringing, to turn a simple "oops" into "knowing infringement".

There are lots of machine learning papers in there, and lots of papers about mobile, search, social networks, etc. This one seemed a little out of place: "Taking Stock of Toilets in Bangalore’s Government Schools: Status, Challenges and Opportunities"

Many if not most large corporations employ anthropologists who conduct ethnographic studies of how people all over the world use technology. Toilet jokes aside, this is not at all out of place.

I had no idea that Microsoft did that kind of research. Maybe this is connected to Bill Gates's philanthropic work.

I don't think so. It's pretty standard for companies with a global presence – they need to employ to find out how (potential) customers use technology in other parts of the world with different cultural and socioeconomic conditions. Intel, IBM, Microsoft, Nokia, etc. all do this. A great blog from a Nokia ethnographer: http://janchipchase.com/themes/future-perfect/.

Also this:

https://github.com/MicrosoftResearchSVC/

Dryad under an Apache license!

See also this more applied site, where you can try out results of some of a part of MS research in your browser -- http://rise4fun.com/ -- this comes from the "research in software engineering" group: http://research.microsoft.com/en-us/groups/rise/default.aspx

Am I the only one who noticed they had articles published in August 2014 on that site?

Also, the first blog post I clicked on [1] brought this quote: "We propose a new progressive analytics system based on a progress model called Prism" WHY is Microsoft so terrible at picking names for their products?

1: http://research.microsoft.com/apps/pubs/default.aspx?id=2001...

It's an MSR publication, not a marketing team pitching a product. I don't think anyone thought hard about the name.

Those are typical names for academic papers, check out pretty much any conference's list of accepted papers for similar verbosity.

> Am I the only one who noticed they had articles published in August 2014 on that site?

It's probably the papers that have just been accepted for publication, but haven't actually been published yet. The submission-reviewing-publication process is very slow and usually takes at least 5 months in CS conferences. So, many authors make the papers available right after they get an acceptance notification from the program committee. So, my best guess is that the publication date refers to the date those papers would be published officially.

Reply on news.ycombinator.com