Yahoo! at CIKM 2008

Napa Valley’s Fall foliage served as the majestic setting for the ACM 17th Conference on Information and Knowledge Management (CIKM 2008), held October 26 to 30. Yahoo! had a great presence, with a total of nine accepted papers and one accepted poster. The conference saw a record number of submissions this year – 772 in all – resulting in an extremely rigorous selection process and an acceptance rate of 17% for papers and 16% for posters. There were twenty-four Yahoos on the program committee, two invited Yahoo! speakers and one tutorial given by Yahoo! Research.

The conference brought together participants from academia and industry, including representatives from University of Washington, Carnegie Mellon University, UMass, Microsoft Research, eBay Labs, Ask.com, LinkedIn, and others.

There were many interesting papers and sessions. Particularly interesting themes of the information retrieval track included evaluation and Web search. Other notable themes were the Semantic Web, filtering, social search, tagging, classification and clustering, and advertising.

The information retrieval keynote was given by UMass Professor Bruce Croft who explained that search is ubiquitous and not just limited to the Web. His examples included patent search, enterprise search, community-based question answering (such as exemplified by Yahoo! Answers), as well as others. He described how long queries are a challenging issue for these types of searches and proposed several ways of dealing with such queries. The other keynotes were given by Rakesh Agrawal of Microsoft Research, who addressed data mining, and University of Washington Professor Pedro Domingo, who talked about Markov logic.

Noteworthy papers included “How Does Clickthrough Data Reflect Retrieval Quality?” by Filip Radlinksi, Madhu Kurup, and Thorsten Joachims. The paper looked at various ways of automatically evaluating search engines using clickthrough data.

Yahoo! Researcher Torsten Suel gave an invited talk at the well-attended 6th Workshop on Large-Scale Distributed Systems for Information Retrieval (LSDS-IR 08), held at the CIKM conference. This was a smaller, specialized half-day workshop on distributed and P2P information retrieval. The first two talks in the session appeared to be extremely popular. The first one, “Robot Army: A Distributed System for the Casual Manipulation of Massive Data Sets,” was about a competitor to Hadoop that is being released open source. The second talk, “Managing Collaborative Feedback Information for Distributed Retrieval,” looked at a peer-to-peer approach to collaborative information retrieval.

Conference-goers also enjoyed a boisterous social program, with a tapas party and a Halloween-themed banquet at COPIA that included a costume contest and a continuous flow of wine from the region.

It was a successful conference overall, and an impressive showcase for Yahoo! and its outstanding accomplishments.

Yahoo! CIKM 2008 papers

"Improved Query Difficulty Prediction for the Web"
Claudia Hauff, Vanessa Murdock, Ricardo Baeza-Yates

"Generalized Inverse Document Frequency"
Donald Metzler

"To Swing or not to Swing: Learning when (not) to Advertise"
Andrei Broder, Massimiliano Ciaramita, Marcus Fontoura, Evgeniy Gabrilovich, Vanja Josifovski, Donald Metzler, Vanessa Murdock, Vassilis Plachouras

"Dr. Searcher and Mr. Browser: A Unified Hyperlink-Click Graph"
Barbara Poblete, Carlos Castillo, Aristides Gionis

"Beyond the Session Timeout: Automatic Hierarchical Segmentation of Search Topics in Query Logs"
Rosie Jones, Kristina Klinkner

"Vanity Fair: Privacy in Querylog Bundles"
Rosie Jones, Ravi Kumar, Bo Pang, Andrew Tomkins

"Search Advertising using Web Relevance Feedback"
Andrei Broder, Peter Ciccolo, Marcus Fontoura, Evgeniy Gabrilovich, Vanja Josifovski, Lance Riedel

"The query-flow graph: model and applications"
Paolo Boldi, Francesco Bonchi, Carlos Castillo, Debora Donato, Aristides Gionis, Sebastiano Vigna

"Supporting Sub-Document Updates and Queries in an Inverted Index"
Vuk Ercegovac, Vanja Josifovski, Ning Li, Mauricio Mediano, Eugene Shekita

Posters

"Search Based Forecasting of Ad Volume in Contextual Advertising"
Xuerui Wang, Andrei Broder, Marcus Fontoura, Vanja Josifovski

"Searching the Wikipedia with contextual information"
Antti Ukkonen, Carlos Castillo, Debora Donato, Aristides Gionis