The Noisy Channel


Upcoming Google NYC Talk: Reconsidering Relevance

January 5th, 2009 · 5 Comments · Uncategorized

I’m givng a talk in the Google New York office this Wednesday (1/7) at 3pm entitled “Reconsidering Relevance”. The title is an allusion to Tefko Saracevic‘s paper, “Relevance Reconsidered“.

I’m not sure how many NYC Googlers read this blog, but I encourage you all to attend. I’ve also been able to put a few folks on the guest list. For everyone else, my host assures me that the recorded presentation will be posted on YouTube. And of course I’ll post the slides here at The Noisy Channel, as well as on SlideShare.

Here’s the title and abstract:

Reconsidering Relevance

We’ve become complacent about relevance. The overwhelming success of web search engines has lulled even information retrieval (IR) researchers to expect only incremental improvements in relevance in the near future. And beyond web search, there are still broad search problems where relevance still feels hopelessly like the pre-Google web.

But even some of the most basic IR questions about relevance are unresolved. ¬†We take for granted the very idea that a computer can determine which documents are relevant to a person’s needs. And we still rely on two-word queries (on average) to communicate a user’s information need. But this approach is a contrivance; in reality, we need to think of information-seeking as a problem of optimizing the communication between people and machines.

We can do better. In fact, there are a variety of ongoing efforts to do so, often under the banners of “interactive information retrieval”, “exploratory search”, and “human computer information retrieval”. In this talk, I’ll discuss these initiatives and how they are helping to move “relevance” beyond today’s outdated assumptions.

5 responses so far ↓

  • 1 jeremy // Jan 5, 2009 at 4:25 pm

    A friend of mine gave a relevance-related talk at Google in the summer of 2008. While giving the talk and chatting with people after the talk, she found that most of the people in the audience had no idea what relevance feedback was.

    Considering that relevance feedback is many decades old, and one of the oldest/most original forms of HCI in IR, I find it a bit disturbing that so many people at this search-focused company were clueless about it.

    I’d be interested in hearing what your experiences are.

  • 2 Daniel Tunkelang // Jan 5, 2009 at 4:48 pm

    Ouch. Though I suspect that Google hires a lot more systems / software engineering generalists than IR or HCI researchers. IMHO, Google’s biggest computer science innovations have been outside of IR and HCI, e.g., Google File System, MapReduce, BigTable. Still, I’m surprised that relevance 101 (in which I’d include both relevance and pseudo-relevance feedback) isn’t part of basic training.

    In any case, thanks for the heads up! I won’t assume that I just race through the basic IR material. Which may be so much for the better, since the online audience will probably be a bit broader than the live one.

  • 3 Patrick Chanezon // Jan 6, 2009 at 3:41 pm

    You should ask your host to setup a Video Conf with Mountain View. I have a meeting at that time, so won’t attend, but I will definitely watch the video!

    For the tech talks I hosted, the ratio of number of video views compared to actual attendees varied from 10 to 5000, so the video is your best vehicle:-)

    Have fun at the NY office.


  • 4 Daniel Tunkelang // Jan 6, 2009 at 4:30 pm

    Patrick, I’ve asked them and hopefully will get an answer in time to post it here.

    Meanwhile, I’m delighted to have you as a reader. I know I can be a tough critic of Google from time to time, but I hope my criticism is fair and constructive. Or at least entertaining.

  • 5 Daniel Tunkelang // Jan 7, 2009 at 10:26 am

    Just in from my host: too late to set up a video conference. But I’ll post the link to the recording on my blog as soon as it’s available.

Clicky Web Analytics