Computer Human Interaction (CHI) is the premier academic conference every year for academic Human Computer Interaction researchers and for professional usability experts, user experience people and interaction designers who are doing cutting edge stuff. Every year, something like 3000 or so people descend for 4 days of often overwhelming talks, panels, demonstrations, video presentations, networking and, ideally, plenty of drinking. It’s considered somewhat of a taste maker and arbiter within this community. No one person can give a completely impartial view of CHI, most of the time, there’s literally about 14 different things going on at once. The most you can do is go to as much of the stuff that interests you as possible and hook into interesting conversation in the crowd.
Still, CHI is somewhat remote from the workaday world of people who are building just another web application or mobile app. The stuff talked about at CHI is often highly abstract and, quite frankly useless to professional developers and designers. Despite that, I love it. It’s a bunch of overwhelming smart people with an astonishingly diverse array of interests all talking about the stuff they’re passionate about with a mix of brilliant insight and hilarious naivety.
Day 3 was a pretty tiring day for me and I was exhausted from the grind and overstimulation of this thing. Most of the talks were uninsipiring and I found myself drifting off a lot but there was one shining bright spot in all of it which was that I went to what I think was the best talk, not only of this CHI, but of any CHI I’ve ever been. So for this post, I’m going to focus primarily on just one talk.
Resonance on the Web: Web Dynamics and Revisitation Patterns
I wasn’t even originally planning to go to this talk. Living in the Seattle area and knowing most of the University of Washington HCI people, my general rule is that I want to reserve CHI time for non-UW talks. However, I had come to that session for an earlier talk I wanted to see and nothing else inspired me so I decided to stay and I was glad I did.
Many of the other talks that I’ve enjoyed at CHI have involved tightly scoping a problem and then coming up with a neat experimental method that probes the question in an effective manner. Eytan Adar’s talk on revisitation and dynamic content was something completely different; it was an exploration into a messy, complex and nuanced data set and an analysis of that dataset which cut through the thicket of noise to produce clear and interesting insight.
“Resonance on the web” is a continuation of some work presented last year at CHI on revisitation patterns on the web. Using an instrumented Microsoft Live toolbar, Eytan collected the usage patterns of 600,000 web users and measured how often they returned to different types of webpages. Some webpages showed short revisitation patterns, most people who return, return 5 minutes or so after their last visit. Others showed that people returned every 6 months or so. By analysing how people revisit a webpage, it was possible to infer what broad class of site it was.
The work presented this year was an extension of this which looks at what people are interested on a page based on revisitation patterns. I’m not going to be able to go into the full subtlety of the talk and I recommend you read the paper for the full details but the basic gist of it is simple:
Most webpages these days are dynamic content. Different elements of the page change, ads change upon every refresh, content changes once in a while, site navigation elements barely ever change. By analysing the same page over time, it’s possible to find which DOM elements change at what rates and then, by matching it to revisitation patterns, you can infer what content people are interested in keeping track of. For news sites like the New York Times, people are constantly refreshing to find the newest content. For shopping sites like Costco, people are mainly coming back to search for things so the navigation bar is the most relevant information. For sites like woot.com, people are mainly interested in the content.
What can you do with this data? Eytan presented two possible applications. One was automatic generation of a mobile site. Since we can infer what content is relevant to most people, we can extract just those elements out for mobile presentation. The other was in generating web snippets for search. Knowing what people are looking for once they’re on the site means you can give them a preview of that information in the summary for search results.
The beauty of this presentation wasn’t just in the results though, it was the entire process that was used to arrive at this understanding and, if there’s one paper you read from CHI this year, I recommend this one.