Social Media Research Blog

Education, Employment and Immigration

2009-08-23T19:44:14-04:00

Techcrunch has a great article talking about the decline of foriegn students in US universities.

"U.S. grad school admissions for would-be international students plummeted this year, according to the Council of Graduate Schools—the first decline in five years. The decline was 3% on average, thanks to increases from China and the Middle East, but some countries saw double-digit declines in interest in a U.S. education. Applicants from India and South Korea fell 12% and 9% respectively—with students turning their sights on schools in Asia and Europe instead.

This shouldn’t be a surprise. Much of the world’s economic growth—hence, jobs—is in emerging markets, the schools are far cheaper and in many cases competitive academically, and then there’s the H-1B issue. If America won’t allow a PhD just trained in our top schools to work here and contribute to the economy—why come here and take on the student loans to begin with?"

Having recently graduated from a US university and having seen the gradual decline in students, this article comes as no surprise. I have some personal experience with this and can say a few words about the process:

The economy in India is doing much better than before. Despite the meltdown and the world wide recession, companies that are focused on the Indian markets are still doing great. Many engineers in India are now finding high paying jobs (unlike the situation in 2001).
The process of obtaining a US Visa -- be it a student visa or a visitors visa is excruciatingly painful and at times downright stupid. Having a visa rejected (at times; without any clear reason) is quite demotivating and many people simply do not really wish to put up with it.
Once you do have the visa, however, things can be somewhat straightforward while you are in grad school. As long as you are a student and your I20 is valid there is little to be worried about. The good part of the process is also the fact that in the United States, you get to do Curricular Practical Training (CPT) and Optional Practical Training (OPT) while you are in school. Most foreign students utilize this to gain experience and skills in the industry. (Although, you land up paying a lot of out-of-state tuition so that you can do this internship!).
The part that starts becoming difficult is the point when your company needs to sponsor for an H1. Many companies exclusively ask that candidates requiring H1s need'nt apply. This year there are still 20K H1s open. Last year, things were much crazier. There were limited number of H1s and too many applicants -- resulting in a lottery system. I personally know enough people who were affected by this madness and had to either reapply or work from Canada/India until the H1 was approved. The basic problem with H1s is:
- The quota/caps on H1s used to get filled very quickly -- up until the admissions to grad schools in US declined drastically.
- They are often abused. Consultants and other contractors file for H1s -- at times for candidates who are not really qualified. Many large companies find it cheaper to hire temporary workforce via contractors -- adding to the problem.
Next, while you are on a student visa (F1) building a startup is extremely complicated. If there was one place in the world where I would imagine this could even be possible would be here in the US. There are ways to do this but all pretty complicated to work out.
Finally, getting a green card requires a wait time of about 5 years (for India) under the EB2 category. Moreover, if you are applying for a green card and get married (to someone who is not on H1/greencard) after your green card is approved it can take forever to get your spouse added to your green card. I know atleast two PhDs who had to leave the country because they could not bring their spouse over.The EB3 category can take 7-9 years!

No system is perfect. Despite the above problems United States is still a place that attracts the best talent in the world. As the world becomes flat and more opportunities/resources become available in other places, keeping the competitive edge for many businesses requires that we reconsider some of the laws around immigration and visas. I really hope that the Obama administration is able to change the legislature to carry out the appropriate reforms.

CFP: 1st International Workshop on Mining Social Media

2009-08-22T20:38:16-04:00

via José Carlos Cortizo Pérez

CALL FOR PAPERS

1st International Workshop on Mining Social Media
http://www.socialgamingplatform.com/msm09/
Workshop at CAEPIA 2009 (http://www.lsi.us.es/caepia09)
November 9, 2009, Sevilla, Spain
Submission deadline: July 31, 2009

OVERVIEW

Social Media are technological tools that allow users sharing and discuss information. Most Social Media are Internet based applications that manage textual information, as blogs (Blogger, Wordpress), microblogging (Twitter, Pownce), wikis (Wikipedia), forums, or Social Networks (Facebook, MySpace, LinkedIn). But there also exist other Social Media Internet applications where users share more than text, as photo sharing tools (Flickr, Picasa), video sharing (YouTube, Vimeo), livecasting (Ustream), or audio and music sharing (last.fm, ccMixter, FreeSound). More recent Social Media includes virtual worlds (Second Life), online gaming (World of Warcraft, WarHammer Online), game sharing (Miniclip.com) and Mobile Social Media like Nomad Social Networks where users share their current position in the Real World.

Social Media have been able to shift the way information is generated and consumed. At first, information was generated by one person and “consumed” by many people, but now the information is generated by many people and consumed by many people, changing the needs in information access and management. It is also noticeable that Social Media applications manage huge quantities of users and data: Facebook and MySpace manage between 100 and 150 million users, it is estimated that 1 million blog posts are generated each day, microblogging services like Twitter generates 3 million messages each day, YouTube manages more that 150.000 million videos, etc. All these points make clear that Social Media is an excellent application field for data miners.

INVITED SPEAKER

* William W. Cohen (Carnegie Mellon University)

INDUSTRY PANEL

The Industry Panel features panelists from several Social Media companies, who will give their vision about needs, opportunities and current solutions of Data Mining and related techniques to real Social Media solutions. Up to date, Strands (http://www.strands.com) and Tuenti (http://www.tuenti.com) have confirmed their presence in this panel.

Further details on the website.

Geolocation and Twitter: Realtime meets Local

2009-08-22T20:33:47-04:00

Recently, twitter announced that it was going to add support for geo annotations for tweets. The original announcement is on twitter blog.

One fascinating thing about Twitter is how the community is involved in shaping the way the product itself evolves. There are several examples of how this happens:

For conversations, the community evolved method of @ replies.
Retweets have become a great way to pass information along and attribute it to the person who provided this information.
Hashtags are becoming even more popular, with almost a handful of hashtags landing up on the twitter homepage everyday.
Location is'nt new either. Many users already provide the location information in their profiles. In addition several formats have evolved where L: is the most popular method for indicating the current location of a user. This was partly popularized by the Twittervision application.

By standardizing the support for geolocation in twitter, the service provides developers with several opportunities to build applications that are geographically relevant. One possibility for twitter is to capitalize on the wealth of user generated geodata:

Think realtime meets local: It would be great for instance, to know that there is a live concert that everyone in my town is talking about right now!
Local Advertising: This could be one direction that Twitter might be heading towards.
Hyperlocal News and Activism: Social media is defining new ways to share hyperlocal news and build community involvement. Twitter is being used by several NGOs, blood banks and Local Police as a means to bring awareness and build campaigns. Applications built using geolocation information can accelerate involvement and social good at a local level.

BTW, here is an interesting stat about twitter users: From a large collection of twitter profile pages I find that roughly 60% of them have some URL associated with the profile. AND of those that have a profile URL, 92% already declare some location for themselves (irrespective of whether the location can be resolved correctly or not). While many worry about the fact that our location privacy is being lost forever; like any technology, it will ultimately be up to the community to come to a consensus on what is acceptable. For most part, I think that users are happy to share their location at some level of general granularity, in exchange of some (perhaps perceived) social value (meeting friends, finding local hangouts or for simply being a bit exhibitionist about what you are currently upto!).

An XKCD Inspired Sketch

2009-08-04T00:47:24-04:00

Akshay Java

Hit me on twitter

2009-04-18T16:46:10-04:00

A shout out to Mistah F.A.B for the new single "Hit me on Twitter"! ROTFL..

Interesting tweetweek with Oprah, Ashton and @CNNbrk making sure that everyone is partying on twitter!

ICWSM Dataset Workshop Accepted Papers

2009-03-20T15:45:07-04:00

We have a great lineup of papers at the main conference and the dataset workshop. Here are the accepted papers for the workshop

Identifying Personal Stories in Millions of Weblog Entries
Andrew Gordon and Reid Swanson

SentiSearch: Exploring Mood on the Web
Sara Sood and Lucy Vasserman

What Bloggers Talk About: Interconnection and Content Sharing in Weblogs
Meeyoung Cha, Juan Antonio Navarro Perez, and Hamed Haddadi

Event Intensity Tracking in Weblog Collections
Viet Ha Thuc, Yelena Mejova, Christopher Harris and Padmini Srinivasan

Quantification of Topic Propagation using Percolation Theory: A study of the ICWSM Network
Ali Azimi Bolourian, Yashar Moshfeghi and C. J. van Rijsbergen

Also a quick reminder -- Today is the last day for early registrations for ICWSM!

SXSW Panel on Collaborative filters

2009-03-15T14:55:44-04:00

I am at SXSW this weekend and wow what a conference this has been! Met a lot of cool people and been to a bunch of happening parties! I plan to post a few updates about the conference. In the mean time here is a quick post from notes I wrote during the panel discussion on Collaborative Filters.

Anton Kast from Digg.com gave a brief history of collaborative filtering from the research literature covering tapestry (from parc) grouplens (UMN) and a few other projects mostly from 1990s and draws comparisons to current systems used in spam filters, pagerank , tagging systems and applications in facebook ads etc. The talk wasnt very technical and meant to be a quick overview for the audience that might not be familiar with the research background. Kast described recommendation as the output of collaborative filtering that is personalized for a particular user in the system. For example amazon (people who bought this also bought..), behavioral ad targets, google news etc. I wish Kast had spent more time discussing the last slide which covered practical problems: sparsity problem in recommendation systems, early rater problem/cold start, gray sheep (small group very popular).

Erik Frey from last.fm shared some interesting stats -- 25m unique users/month. The types of relations that last.fm derives: songs-users, users-users. He also shared some insights on what were described as lean forward and lean backward mode in music recommendation systems. The lean forward mode is when a user is feeling a little experimental and would like to discover new stuff. The lean backwards mode is when the emphasis is on continuity and the desire to listen to something familiar (i.e. radio experience). The two main data sources for last.fm are : scrobbles, social info like tags. Frey suggests that tags are quite useful to explain recos

The next speaker was the cto of baynote -- a recommendation system that is provided as service using javascripts that enables clients to customize their website in real time based on user behavior. They also apply it for reordering the search results.

Finally, Jon Sanders from Netflix gave a very good talk covering the history of netflix and its evolution. Today netflix has about 2 billion ratings and gets about 2 M ratings per day. 60% of movies watched are via personalized recos -- amazing!

Following were the steps in evolution of netflix:

step 1 At first netflix was a editorially managed site and the first evolution was that of adding a rating widget.

step 2 score and sort all movies using rating alone (top k list)

step 3 movie similarity by k-nearest

step 4 interest based recos (same actor director etc)

step 5 ask ( get users to rate genre)

step 6 ask others (top 10 lists from users)

step 7 explain why (because you enjoyed)

step 8 $1M prize

step 9 develop a unique experience and customized website for each customer

future: streaming video specific recommendations, incorporating implicit and explicit metadata, discovery focus, let people drive and not be led.

Microsoft TechFest 2009

2009-02-26T16:19:27-05:00

This was my first experience at Microsoft's annual TechFest Event

"TechFest is an annual event that brings researchers from Microsoft Research’s labs around the world to Redmond to share their latest work with Microsoft product teams. Attendees experience some of the freshest, most innovative technologies emerging from Microsoft’s research efforts. The event provides a forum in which product teams and researchers can discuss the novel work occurring in the labs, thereby encouraging effective technology transfer into Microsoft products."

It was an opportunity to totally geek out on some really cool technology and an an absolutely fantastic chance to peek into what the future of computing holds for us. I am simply amazed at the amount of cool stuff that is out there. Ofcourse, I cant talk about many of the projects on display at TechFest 2009 but here are a few public videos you should definitely check out! A quick shout out for GeoLife 2.0 and Social Desktop among the ones (publicly available) that might be of interest to the social media community:

TechFest: Social Desktop

TechFest: GeoLife 2.0

Also check out all the buzz around techfest on Twitter!

Freedom of Speech Blogging

2009-02-25T15:36:23-05:00

This is a really disturbing case and I am really infuriated by the Indian SC's politically biased decision in this matter:

"This chilling warning emerged as a Bench comprising Chief Justice K G Balakrishnan and Justice P Sathasivam refused to protect a 19-year-old Kerala boy, who had started a community on Orkut against Shiv Sena, from protection against summons received from a Maharashtra court on a criminal case filed against him.

Petitioner Ajith D had started a community on Orkut against Shiv Sena. In this community, there were several posts and discussions by anonymous persons who alleged that Shiv Sena was trying to divide the country on region and caste basis.

Reacting to these posts, the Shiv Sena youth wing's state secretary registered a criminal complaint at Thane police station in August 2008 based on which FIR was registered against Ajith under Sections 506 and 295A pertaining to hurting public sentiment."

Shiv Sena -- a politico-religious party in India is often blamed for extremist views and many people have been outspoken in their criticism about Shiv Sena's policies. I am just shocked that this case would even have to be appealed in the Supreme Court of India and even worse -- the SC of India refuses to protect it's citizens freedom of speech. Ajith's act of starting an Orkut community should not matter. The views expressed in the community are those of individual citizens --free to exercise their fundamental right to speech. The fact that it is in a new, social medium rather than a rally or even printing pamphlets does not make it a criminal act. There is nothing wrong about what Ajith has done and it is far from being criminal in any way. Some people may not agree with his views or the views of members of this community. Isn't that the fundamental right we all enjoy in a free society? How would India be any different from China or Iran if the voices of her own citizens are criminalized and persecuted for expressing their opinions?

I call on all bloggers (and especially, India citizens) to take up this case and actively blog about it. It isnt just a question of freedom of speech of a single citizen in some other country but a question of what constitutes the "Freedom of Blogging".

ICWSM 09 Dataset Deadline Reminder

2009-02-03T02:51:02-05:00

A quick note: So far we have had about 185 Dataset downloads!!

The ICWSM-09 Data Challenge Workshop will be held May 20, 2009 in San Jose, California in conjunction with the Third AAAI International Conference on Weblogs and Social Media. Submissions for the workshop are due March 1, 2009. Please see http://www.icwsm.org/2009/data/ for complete submission details.

The ICWSM-09 Spinn3r blog dataset is a collection of 44 million blog posts made between August 1st and October 1st, 2008, and collected by Spinn3r.com. This dataset is freely available to researchers under a liberal data usage agreement.

Authors are invited to submit papers to a data challenge workshop to be held on the last day of ICWSM-09. This workshop will feature research papers as well as a wide-ranging discussion of data issues facing the social media research community. Good research topics might include:

* link analysis;
* social network extraction;

* clustering and topic identification;
* tracing the evolution of news;
* blog search and filtering;
* psychological, sociological, ethnographic, or personality-based studies;
* analysis of influence among bloggers;
* blog summarization and discourse analysis.

You should feel free to explore any aspect of the data that you feel would be of interest to the ICWSM community. An award will be presented at ICWSM for the best paper using the dataset.

Papers may be submitted online at http://www.easychair.org/conferences/?conf=icwsm09dcw.

Submissions may be up to 8 pages in length, must be in PDF format, and must follow the ICWSM formatting guidelines.

Best regards,

Ian Soboroff (NIST)

Akshay Java (Live Labs, Microsoft)

ICWSM-09 Data Chairs