UMBC ebiquity
2006 March

Archive for March, 2006

SMS banking

March 31st, 2006, by Anand, posted in Uncategorized

I recently came across this site offering a payment service over the phone. It allows regsitered users to transfer small amounts of money back and forth by sending text messages of the form “pay 5 111-222-3333.” A callback from their service then requests authorization — entering a secret pin-code. SMS banking? Pretty neat idea, couldn’t banks do this too (with specified upper bounds for all transactions)?.  Here is the site:
TextPayMe

This convenience comes with softer security guarantees, SMS transmissions are insecure, and so is the pin-code process. But if the amounts are trivial — the ease of use is compelling!

BabyWear is like an ipod for babies

March 31st, 2006, by Tim Finin, posted in Uncategorized

The UMBC wearable computing group led by Professor Zary Segall is developing human-aware technology. One recent prototype is BabyWear, a system designed to connect premature babies in neonatal intensive care to their mothers via sound. The system is designed to reduce the stress on both infant and mother and minimize the time in neonatal intensive care. Here is a short video report done as part of CNN’s Pioneers series.

AAAI Weblogs Symposium – Panel – Day1

March 27th, 2006, by Pranam Kolari, posted in Uncategorized

A panel on “Technologies to Understand it Now and Gain Insight in the Future” was part of the AAAI Weblogs Symposium today.

Panel Technologies to Understand it Now and Gain Insight in the Future
  Moderator Mark Liberman
  Participants Chris Redlitz (President, Feedster)
  Michael Sippey (Vice President Product, Six Apart)
  Howard Kaushansky (CEO, Umbria Inc.)
  Tony Perkins (Founder and Creator, AlwaysOn Network)
  Andrew Bernstein (CEO, Cymfony)
  Carrie Grimes (Group Lead – Search Quality, Google Inc.)
  Cameron Marlow (Yahoo Research Berkeley)

The panel was organized around questions. We summarize what the panel thought about these questions below. Panelists are identified by their abbreviated names.

What information do you get from Blogs? HK’s take is “Market Research”. Putting the “finger on the pulse” of consumers. AB agrees with HK and also points PR Marketing. AB discusses blog analysis in addition to e-mails received by organizations, and message-board discussions and the task of correlating all of them, with blogs as the pivot! AB mentions splogs and how they are effecting their analysis. CR talks about growth of the blogosphere. It all started as “The most recent post about X”. Relevancy is becoming more important now. CR talks about AOL connection and how AOL is using their index to track conversations on the blogopshere. CR ends by saying that traditional media is now open to involving unedited content on their pages.

How good is it? How much does it matter? Discussion was centered around SPAM and related issues. The question is regarding the robustness of current analysis techniques against spam? CG says don’t worry about it – data is always dirty. Blogger has cracked down spam blog postings. CG deviates from the problem and suggest that query disambiguation and other issues are more important. HK brings back the topic by asking about conflict between search engine revenues and SPAM. MS talks about how spammers go to such an extent as creating paid accounts on TypePad and how Six Apart does not allow automated content generation. MS says blogosphere does care about splogs as opposed to CG. ML gives a great example where hijacked content from his blog listed on another site had outlinks replaced to porn sites.

How will consumers use these analyses? CM raises an interesting point of privacy being an issue in the next 5 years, as organizations increasingly generate blog data. IP rights for RSS feeds is another issue. MS says we will see a bifurcation between bloggers that want to be public and bloggers who don’t. MS also talks about privacy issues in the future. Some bloggers say — “I don’t want to be in Google’s index”, but want to talk about it with my friends. MS says analysis engines have to worry about not having access to this data in the future. TP says privacy won’t be that important in the next generation, he says go look at “MySpace”, and says that tools should look at integrating blogging with social networking.

What do you need from researchers? CG gives a broad view, and talks about personalization in general. MS says there are 3 million active LJ users and there are many coommunities. MS points to recommending communities as very important from their perspective. AB talks about having better tools for relevancy. CR supports the community view. TP talks about employee group blog and how employees can make the entire organization more competitive. HK says there isn’t anything we cannot do with language technologies but almost all of them are not sufficiently accurate. So he suggests that researchers should work on making them more accurate.

Computational Approaches to Analyzing Weblogs

March 27th, 2006, by Pranam Kolari, posted in Uncategorized

AAAI Spring Symposium started today at Stanford. StanfordWe have a presentation on Wednesday, March 29th, based on a part of our work on splog detection. We are always open to discussion on any topic related to the Blogosphere or the Semantic Web. Catch any of us around — Tim Finin, Pranam Kolari or Akshay Java — anytime!

Google Techtalks

March 26th, 2006, by Amit, posted in Uncategorized

Some of the amazing technical videos from “Google techtalks” that take place at the Mountain View Campus!

The videos are huge, about an hour each, but really a treat for the brain with amazing ideas. Its all geekie stuff …

http://video.google.com/videosearch?q=techtalks&page=1&lv=0&so=1

Its a must watch for all according to their tastes.

Google to patent ads localized to wifi access points

March 24th, 2006, by Tim Finin, posted in Uncategorized

Google’s plans for supporting metropolitan wifi access points in major cities are becomming a bit clearer. Google has applied for a patent on a technique to show wifi users ads based on the location of the wireless access point they are using. Such ad supported wifi hotspots can be free or heavily discounted. Google might also be able to draw on additional information it has about users to further refine the ad selection. Kevin Newcomb writes

One benefit put forth in the application is to give wireless Internet providers an ad-supported business model to offer free or discounted Wi-Fi access to users. This would help overcome the “the gap between what Wi-Fi operators charge and what casual mobile users are typically willing to pay,” according to the document. … The document describes applications of the patent where ads would be shown in a toolbar, on a Web page, or during page transitions. It also specifies that ads would continue to be served regardless of whether a Web page, or other document being viewed is updated.

This ties in rather nicely with Google GeoAds that appear on Google local searches.

Most discussed blog post of all time

March 15th, 2006, by Pranam Kolari, posted in Uncategorized

A question occured to me when going through Tim Berners-Lee’s blog. TimBL’s first post has 455 comments (at which point comments were turned off) and 200 citations/inlinks (make that 201!) per Technorati. I could not figure the top posts for any of Steve Rubel’s, Scoble’s or Om Malik’s blogs — tech blogs I regularly follow, and possible candidates for the top cited post of all time. So here’s my primary question:

What is the most discussed blog post of all time — comments, inlinks combined?

Understandably, blog search engines are still not capable of indexing and analyzing comments. With the current capabilities, listing the most cited blog posts of all time (on a single page) based on inlinks would surely be feasible, and interesting. Top posts aggregated over authors would be a great complement. Though some might argue that these features are not all that important for the blogosphere, it will still help in understanding what makes the best and most useful blog posts, similar to how we (graduate students!) in academics use top cited papers, both as an insipiration and a guide to work on topics that make a difference. BlogPulse seems to have something close — top blogs for the day — but nothing aggregating them over time or author. Any other answers?

VeriChip targets DC market for its RFID implants

March 15th, 2006, by Tim Finin, posted in Uncategorized

The Washington Post has an article that describes how VeriChip is marketing it’s implant RFID chips to the DC metropolitan area.

Use of Implanted Patient-Data Chips Stirs Debate on Medicine vs. Privacy … “The two D.C. residents are among just a handful of Americans who have had the tiny electronic VeriChip inserted since the government approved it two years ago. But the chip is being aggressively marketed by its manufacturer, which is targeting Washington to be the first metropolitan area with multiple hospitals equipped to read the device, a persuasive factor for Fischer and Hickey. Within weeks, the first hospital is expected to announce plans to start routinely scanning all emergency-room patients.”

It is going to be an uphill battle to convince people to do this.

“The company has sold about 2,500 chips worldwide for use in people, and several hundred have been implanted, including about 100 in the United States, spokesman John Procter said. So far in the United States, however, most of the chips have been implanted into the company’s own employees. Suspecting that many people are hesitant to get the chips until more emergency rooms are able to scan them, the company has begun giving scanners to hospitals for free, Procter said.”

Two issues that should make us wary are privacy and security. The privacy issues are well known, but here is also a serious security problem. VeriChip is also marketing implanted RFID tags for controlling access to secure or sensitive areas.

The company is, however, marketing the devices to limit entry to secure facilities. The Mexican government is using the implants like key cards for high-security offices. And CityWatcher.com of Cincinnati, which stores surveillance-camera footage from around the country, recently started using the chips to control access to tapes. Bars in Spain and Amsterdam, meanwhile, are offering the chips to patrons who want quick entry and to run electronic tabs.

But, it turns out to be easy to skim and clone a verichip, as cq.cx demonstrates:

… “Verichip markets their product for access control. This means that you could have a chip implanted, and then your front door would unlock when your shoulder got close to the reader. Let us imagine that you did this; then, I could sit next to you on the subway, and read your chip’s ID. At this point I can break in to your house, by replaying that ID. So now you have to change your ID; but as far as I know, you cannot do this without surgery. …”

RDF/A Primer released

March 14th, 2006, by Tim Finin, posted in Uncategorized

The W3C’s HTML and Semantic Web Best Practices and Deployment Working Groups have published the First Public Working Draft of the RDF/A Primer 1.0. RDF/A provides a set of attributes to embed RDF in XHTML documents and supports embedding RDF without repeating existing XHTML content when that content is the metadata, as in the following example:


<h1 property=”dc:title”>Vacation in the South of France</h1>
<h2>created
   by <span property=”dc:creator”>Mark Birbeck</span>
   on <span property=”dc:date” type=”xsd:date” content=”2006-01-02″>
   January 2nd, 2006
   </span>
</h2>

Eliminating Duplicate Answers in Prolog

March 12th, 2006, by Tim Finin, posted in Uncategorized

I was recently asked

Say I have the sister relation defined like this:

sister(X,Y) :- female(X),parent(P,X),parent(P,Y), + X=Y.

And say I have a DB full pairs that Prolog can figure out are sisters. If I try to find all the sisters using a query like this:

?- sister(X,Y).

I get each pair in each order. Is there any way to make the query such that all of the pairs of sisters are listed only once, in whichever order Prolog finds first?

This is a classic problem that’s also common in a databases query context (SQL is also very declarative). The problem gets right to the issue of whether we want to program declaratively or procedurally. Additional complications are introduced by the desire to write Prolog predicates that can serve either to prove that a relation holds between values or to generate values for which a given relation holds. I tried to fire off a quick and hopefully helpful response to the question, but it spun out of control. To quote Piet Hein, “Problems worthy. of attack. prove their worth. by hitting back.” After working to write it up, I thought I’d put it online in case others might find it useful.

The risk of just tearing up unwanted credit card applications

March 11th, 2006, by Tim Finin, posted in Uncategorized

In The Torn-Up Credit Card Application, a prankster describes how he did everything he could to make his (unsolicited) Chase credit card application look suspicious. Here’s how Chase recommends disposing of unwanted applications:

“If you receive financial solicitations that you’re not interested in, tear them up before throwing them away, so thieves can’t use them to assume your identity.” source

He tore the application into small pieces; taped them back together; specified that the card be shipped to a different address than the one the application was sent to; and gave a cell phone number as his contact. He got the card, of course. Scary.

DOD and Wal-Mart troll the blogosphere

March 7th, 2006, by Tim Finin, posted in Uncategorized

We’re working on a project that models how ideas and influence spreads through the blogosphere. One use for the model might be to find a small set of well connected blogs such that influencing them might eventually lead to a tipping point in which a large fraction of bloggers were influenced.

If you mention certain small technology companies or their products in your blog you might be contacted by them, or at least find relevant comments on your posts. That seems very natural. We’ve got Google alerts set to notify us if others are talking about our research group. We want to know what people are saying and thinking about us. This week I was surprised by two examples of big organizations trying to monitor the blogosphere and influence the bloggers.

The U.S. Central Command, which runs military activities in Iraq and Afghanistan, has a three-man team to interact with bloggers who post about relevant topics and offer them information. This is from an article from the American Forces Press Service:

McNorton said the team contacts bloggers to inform the writers about any given topic that may have been posted on their site. This outreach effort enables the team to offer complete information to bloggers by inviting them to visit CENTCOM’s Web site for news releases, data or imagery.

The team engages bloggers who are posting inaccurate or untrue information, as well as bloggers who are posting incomplete information. They extend a friendly invitation to all bloggers to visit the command’s Web site.

Many bloggers appreciate the team’s contact, blog team officials said, and most post CENTCOM’s Web site as a link on their blog sites. This, McNorton said, has a “viral effect” that drives Internet news consumers to CENTCOM’s Web site.

“Now (online readers) have the opportunity to read positive stories. At least the public can go there and see the whole story. The public wants to hear these good stories,” he said, adding that the news stories the military generates are “very factual.”

It turns out that Wal-Mart has also been trolling the blogosphere and trying to plant good news stories. See, for example, this NYT article about Wal-mart’s efforts to influence friendly bloggers.

Under assault as never before, Wal-Mart is increasingly looking beyond the mainstream media and working directly with bloggers, feeding them exclusive nuggets of news, suggesting topics for postings and even inviting them to visit its corporate headquarters.

You are currently browsing the UMBC ebiquity weblog archives for March, 2006.

  Home | Archive | Login | Feed