February 26,2009 by gleicher (link)
Please post comments after the talk

Everyone should post at least one comment.

March 2,2009 by tgrim (link)
Definitely an interesting talk.

The brunt of it was statistical analysis of websites over time and information available. It was focused mostly on the Zoetrope and Ziggurat technologies that he had worked on. There didn't appear to be much to the visualization, short of a graphs and plots of data derived from the websites. I was hoping to see something about GUESS or some of the other visualization systems he had helped with, but it wasn't in the scope of this talk.

March 2,2009 by mccardel (link)
talk response

Before the talk, me and some other guys were wondering what it had to do with graphics, and we figured out that it was partly a problem of visualization. I was looking for that sort of stuff, and didn't see too many fancy things related to that topic. However, I still found the talk pretty interesting and the idea behind it was cool.

March 2,2009 by aderhold (link)
Technical content comment

I found much of the candidate's talk to be quite interesting. As someone who will be working for a large web company after I graduate this spring, I think I could potentially incorporate some of his techniques for gathering user visit and revisit data into any design decisions that I might have to make in my own career. I also wish I could download a copy of his Zoetrope program for my own computer. That would be really cool.

March 2,2009 by finn (link)

I'm not sure what to say, but I guess I found it somewhat interesting.
It sounds like an implementation and maintenance nightmare. His example of tracking the price of top-item lists on amazon was interesting, but did he ever explicitly answer Suman's question about tracking the price of a specific item? Is that where "binding" two lenses comes into play? I forget. Anyway, I'm not entirely convinced that this will be of real use to the average user, except for perhaps the mobile device example and price tracking. Even then, someone brought up the point that amazon might not like the idea of someone tracking prices. For that matter, would google really want "better" search snippets, such as important price information? Wouldn't that possibly cannibalize their sponsored links? Like I said, it's a cool idea, but I don't really see where I specifically would make use of it.

March 2,2009 by alex (link)
...

Relatively interesting talk about harsnessing the dynamic nature of the internet. I was particularly impressed by the different lenses in Zeotrope for scanning through a page's updates section-by-section. Maybe it's not technically the most challenging thing, but I think as an idea it's pretty good, and can see browsers in the future incorporating some flavor of that functionality.

March 2,2009 by rosin (link)
no subject

The most interesting part for me was the possible application to improve the blurb on search engine results. I wonder how reliable something like that would be in practice - the idea of trying to find a synopsis most relevant to what the user probably wants, rather than most relevant to the search query, actually reminded me of Cuil's failed launch. If this sort of dynamic blurb/thumbnail selection system works in the general case I'd be very impressed.

@finn: I think that's the purpose of the lens that looks for similar textual layout (tracking scores for the same team over multiple games was the example). Then again, that lens outline was green, and the one he showed for Amazon prices was orange/red.

March 2,2009 by elisabet (link)

I missed the talk because of my 302 lab, but I read the paper. It was pretty interesting, most of it was a high-level explanation, but it made sense. There was a lot about how to make the program user-friendly and the visualizations smooth despite changes in websites over time like features jumping around. Their visual goal seemed to be stability despite their data's general lack of that.

March 2,2009 by yuzhen (link)
Two applications I like

I don't know much about the technology talked today, but I found two interesting applications.

The first one is utilizing history data to view the DVD prices over a long time. Before I came to Madison, I wanted to know how the weather changes from spring to winter.

The second one is viewing web pages on cell phone. He use temporal information to choose the potential interesting / important parts to user.

March 2,2009 by yangk (link)

The presentation was somewhat interesting. I can see how Zoetrope and Ziggurat could be useful for some. The rectangle box lens just looked cool. It was interesting how they used the data they collected to try to figure out what content the average user would be most interested in. Being able to compare the price of an item by checking its history would be a very useful tool since I buy most of my electronics online.

March 2,2009 by zoerb (link)
Zoetrope Paper

I couldn't go to the talk because of a conflict, but I read the Zoetrope paper. Zoetrope is basically a program for indexing and viewing websites as they were at various times in the past. The author's crawler implementation indexes a chosen subset of webpages every hour and stores both XML data and an image of what the webpage looked like at that time. When needed, a user can view this webpage history by using what the authors call Temporal Lenses. These are UI boxes that can be drawn around a particular element of a webpage, and have the option to filter what the lens shows the user by time. The user can interact with the old webpage (as seen through the lens) just as they would the current webpage.

March 2,2009 by amoore (link)
Eytan Adar

This was about viewing web pages over time. The pages are crawled and collected in a database. Then they can be viewed in the future by setting up windows that can be scrolled through time. I think the windows were one of the best features of Zoetrope. If you could set up those windows on your computer and access them on an iphone, you could quickly find useful info. This seems to be really good if you could crawl everything you wanted to know before you knew you needed it, but because the size of your hard drive and speed of your connection are limited, this may not be always the case. Perhaps large news sites could keep their own databases of previous pages so we do not all need to keep our own.

March 3,2009 by cory (link)
Technical Content

I found the talk to be interesting. His methods for analyzing how a website was change were interesting, and I found it cool that these changes stablize at one point. It makes a lot of sense if you think about it, but it's not something I'd ever thought about before.

Like others I question the implementation. It would require significant storage space i would think and he didnt cover whether that would be locally stored or server based. I could see Zoetrope being using for reverting webpages though.

Ziggurat was a nifty idea as well, but i feel it could be very annoying if vandals (*cough* 4chan *cough*) raid a wikipedia page. Suddenly their vandalism would be suggested to everyone, and possibly pushed out by people that dont get the joke or reference.

I'd also like to see how fast Zoetrope would gobble disk space tracking an extremely hightraffic page that any of the megalith internet communities inhabit. Image boards especially.

EDIT: Adding on, I felt that Zoetrope seemed like it might have a great amount of application for marketing more so than the average user. It reminded me of something Professor Miller said in OS about how there is a paradigm shift occuring where there is a demand for higher level programming languages. Zoetrope struck me as one of these higher level language tools. Users could use intuitive systems generate and analyze data, and in that regard, I could see it having a lot of application even outside the web, say on spreadsheets. Highlighting boxs and clicking to activate links is a lot easier on the average cube-slave than the current sorts of programming back end required, and in that sense, it might also have a lot of use in the business and statisics world if they exported it for other uses (or the smart business sets their data up on a webpage on an isolated intranet).

March 3,2009 by cory (link)
On the talk itself

The talk itself was interetesting. If i'm not mistaken, this talk was basically Eytan showing off his research and telling the CS department why we want to hire him?

Either way, it was interesting to see some of the audience's reactions. When he began talking about the change curves of the websites, I definately saw Prof. Rom's ears prick up and he began nodding along or frowning at the various things Eytan had to say.

March 3,2009 by jhugo (link)

@cory He was invited to talk about his work, but he's not being considered for hire based on this talk. Many professors are invited to talk, usually one or two a month from what I remember.

I think the work was intresting, but I didn't see much in results or review of the data beyond what seems obvious. He kept saying people would check Woot every day or so. To me, that seems obvious based on what Woot is. Certainly he implied that on a mobile device, the average user might only need to see the deal of the day, but then he wavered back and said that might not work. I felt like he just made that observation as he was writing his speech, and not as part of his research.

So, interesting work, I'm just wondering how/where it will be used.

March 3,2009 by mikola (link)

The talk was pretty good and I can't complain about free cookies. I know next to nothing about web/HCI stuff, so it is always interesting to get a broader perspective. I was most impressed by the quality of the software the speaker presented; the examples were really sharp and polished.