Friday, April 17

TREC 2009 Web Track details

The guidelines for the TREC 2009 Web Track are available.

There are two tasks:

1) Adhoc -- 50 short 'title only' topics
2) Diversity -- same topics as Adhoc, but:
The goal of the diversity task is to return a ranked list of pages that together provide complete coverage for a query, while avoiding excessive redundancy in the result list. For this task, the probability of relevance of a document is conditioned on the documents that appear before it in the result list.

LingPipe 3.8.0 Major Release

LingPipe today announced that release 3.8.0 is available (details on the homepage).

There is a slew of new features that you will want to check out. Here a few highlights:
  • Google-like auto-complete in spell.AutoCompleter
  • A new, very fast implementation of a classical Naive Bayes text classifier in classify.TradNaiveBayes with support for EM optimization.
  • Overhauled K-Means clustering implementation
  • Updated language features: better use of generics, annotations, etc...
  • Many others...
LingPipe is my favorite library for fast and efficient implementations of core algorithms that I use often.

Wednesday, April 15

Grafitter wins Smiley Award

CMU announced (press release) the 2009 winner of the Smiley Award. The award went to Ian Li, for creating Grafitter. It allows you to create tags that record your behavior, mood, etc... which you send to Grafitter via Twitter, IM, or blogging for analysis and vizualization.

Another project EyeTable, an intelligent restaurant table that uses headsets and sensing technology based on the Wii game controller to determine how well people are responding to one another on dates by analyzing their gestures and speech patterns received an honorable mention.

Dinner Impossible: Yahoo! Search edition

Two of my favorite topics: cooking and search meet tonight. The Yahoo! search blog announced that the Dinner Impossible Food Network show will air an episode tonight that features cooking for the fifth anniversary of Yahoo! search filmed back in February.
Their mission: make the top 15 most searched dishes, each dish using a randomly assigned top searched ingredient.
They even put some of the Yahoo! team to work in the kitchen. Sounds like fun! It's definitely in my Tivo queue.

They posted some photos of the event.

Monday, April 13

ECIR 2009 Best Paper Awards

The European Conference for Information Retrieval (ECIR 2009) was held last week in Toulouse.

Xing brought word that the paper, Mean-Variance Analysis: A New Document Ranking Theory in Information Retrieval by Jun Wang won the best paper award. It applies statistical theory from selecting stocks for a financial portfolio to ranking web documents.

Unfortunately, the proceedings are published by Springer and there doesn't seem to be a copy available online yet.

The best student paper was Joint Ranking for Multilingual Web Search by Wei GAO.

Update: Readers posted in the comments that the best poster award went to A Practitioner’s Guide for Static Index Pruning by Ismail Sengor Altingovde, R. Ozcan, Ö. Ulusoy.