Jenerous is a site that that tries to help “leading marketers, entrepreneurs and innovators jenerously share their stories and knowledge with the world” by publishing podcast interviews. Last week they interviewed ebiquity PhD student Akshay Java and put up a podcast on UMBC’s social media research, including spam detection, topic identification, and trust and influence modeling.
Archive for April, 2007
Yesterday saw another New York Times article on mobile-phone based social networking systems, Social Networking Leaves Confines of the Computer.
“The social networking phenomenon is leaving the confines of the personal computer. Powerful new mobile devices are allowing people to send round-the-clock updates about their vacations, their moods or their latest haircut. New online services, with names like Twitter, Radar and Jaiku, hope people will use their ever-present gadget to share (or, inevitably, to overshare) the details of their lives in the same way they have become accustomed to doing on Web sites like MySpace.”
Also mentioned was Kyte, a service that lets you “produce your own shows and broadcast them to your personal website, blog, MySpace profile or to any of our supported phones”.
Many in the developed countries are skeptical about these microblogging services that are (largely) desgined for mobile phones. Why should I put up with their limitations when I’ve virtually never separated from a good computer with a broadband connection to the internet? These serviced might play an important role in the developing world.
In sub-Saharan Africa, for example, internet access is still very limited but mobile phone penetration is very high. A recent Economist article, for example, said
“Mobile-phone use in sub-Saharan Africa is soaring (see chart). Whereas only 10% of the population had network coverage in 1999, today more than 60% have it, a figure expected to exceed 85% in 2010, according to the GSM Association, an industry trade group.
Maybe the real benefit for microblogging systems designed to work with mobile phones (e.g., Twitter) has yet to be revealed.
Two internal events have provided us with a chance to showcase our research. This past Friday, many of our students participated in the Graduate Research Conference and presented their work. Three Cheers for Anubhav Kale, who won the first prize in the oral presentation session. Nimish Vartak from Ebiquity had won this prize last year. Like Nimish, Anubhav too is headed to Microsoft later this summer. In fact, it seems that all of our graduating MS students are going to Microsoft! Hmmm….
This coming Friday (May 4th) will be the CSEE Research Review day — we will again be represented with talks and posters. The Three Cheers this time will be for Sandor Dornbush, whose work has been judged “Best Research by a PhD student”. This promises to be an interesting event, so join us!
This is the first of a three part sequence. Well, saying it’s about CEO blogs is neither accurate nor fair But we all see examples of blogs that are ostensibly written by people with important and influential positions — politicians, CEOs, Government leaders, division heads — and know that the content is generated by their staff. Click on the image to see the rest on the Dilbert site.
Yahoo! Research has releases an open source version of Pig, an “infrastructure to support ad-hoc analysis of very large data sets” and to do it using massively parallel processing on clusters.
We are creating infrastructure to support ad-hoc analysis of very large data sets. Parallel processing is the name of the game. Our system runs on a cluster computing architecture, on top of which sit several layers of abstraction that ultimately bring the power of parallel computing into the hands of ordinary users. The layers in between automatically translate user queries into efficient parallel evaluation plans, and orchestrate their execution on the raw cluster hardware.
This post does not report any new work. Rather this is a piece of work looking for coders.
I personally don’t care much for the semantic web because the performance of reasoning even on simple ontologies is often very bad. I am currious to see if the performance could be improved if someone parallelized a semantic reasoner like Jena and ran it on a grid like Bluegrit. Deepblue showed that a parallel architecture can be very effective at graph search problems. Usually reasoning is approached as a graph search problem, so can we create a super fast semantic reasoner on our grid?
DBpedia is an interesting project out of Leipzig and FU Berlin to extract structured information from Wikipedia and make it available on the Web and via SPARQL queries. See this ESWC 2007 paper (What have Innsbruck and Leipzig in common?) and slides for an upcoming talk in the WWW2007 developers track. Michael Bergman has an extended post about the project and its possibilities.
Today’s New York Times has an article, Sun and IBM to Offer New Class of High-End Servers, on new specialized high-end servers recently announced by both Sun and IBM. While Sun’s new machines are aimed at managing high-volumes streams of data (e.g., video), IBM’s is designed to support MMORPG games and virtual worlds.
“I.B.M. said its new video game server, which it is calling a â€œgameframe,â€ is being designed in collaboration with Hoplon Infotainment, a Brazilian game developer interested in creating a software layer it calls a â€œbitverseâ€ to support virtual online worlds.
There are already multiplayer games that support hundreds of thousands of simultaneous players, but the I.B.M. system will add an unparalleled level of realism to visual interactions, Mr. Meyerson said.
He argued that in addition to gaming applications, this kind of technology could be used to enhance the performance and scalability of existing virtual worlds like Second Life, an Internet-based service that crosses the boundary between online entertainment and workplace collaboration.”
Hoplon Infotainment runs its Taikodom, a space-based, sci-fi massive multiplayer online game, on conventional IBM Z-series servers. The new IBM “gameframe” machines will combine high-performance features of mainframe computers with the parallel processing capabilities of IBM’s Cell Broadband Engine chip currently being used in the Sony PlayStation 3 game console.
The computer game industry needs software engineers that understand how to take advantage of the new classes of parallel processors that are being developed. As part of our effort to support the GAIM tracks, we plan to offer CMSC483 (Parallel and Distributed Processing) every year.
It is interesting that TBL mentioned the Xanadu project and a reference  in a discussion on handling copyright issues on the Web. The Xanadu project was well known as a pre-web hypertext system, and its infrastructure enables data integration and data provenance annotation.
The first public uses of “hypertext” may be attributed to Ted Nelson in 1965 [http://xanadu.com/XUarchive/].
After that, the project Xanadu  was described in several notes in 1960s.
Xanadu organizes content (e.g. content, comments, revisions) as a network and offers the various UI designs (linked, expandable, zooming, hottext, dissection) for browsing such organized data.
 A presentation overviews Xanadu by comparing it with the Web. http://dc-mrg.english.ucsb.edu/conference/CNCSC/multimedia/documents/wardrip-fruin.pdf
 Xanadu homepage, http://xanadu.com/
I had the opportunity to present my masters work at Vehicle Technology Conference(VTC) Spring 2007 in Dublin. The conference was very interesting but not particularly focused on vehicles. The conference should more accurately described as wireless technology conference. Most of the work focuses on wireless communication at the physical and protocol level. Some of the most interesting posters dealt with communication using white LED’s and human skin.
There were a couple of other interesting papers on VANET’s applied to parking spaces, or traffic lights. The traffic light work focused on using VANET’s to see how many cars at a traffic light. While this is interesting it may not actually work much better than standard traffic light sensors. I would like to see VANET’s applied to agent based traffic lights such as this cool applet.
I have thought for a while about using VANET’s in an auction for traffic lights. In such a scheme every driver could place a value on not waiting at the light. The light would switch when the amount offered by all of the drivers waiting is greater than the offer of all of the drivers approaching the intersection on the other road. In this way one could pay for faster access to roads. Such a system need a safety valve so no person would end up in a starvation state never making it through a light.
I was pleasantly surprised when an author of a semantic web-service paper was asked if they used Swoogle. It is nice to know it was known by both the author and the audience. The author indicated that they would like to link to Swoogle in the future.
VTC is coming to Baltimore in the spring. They are looking for a local liaison and volunteers. Anybody interested in wireless networking and have some free time?