Some Highlights of the CIKM 2011 Conference

riclage's picture


From October 24 to 28, I attended the 20th ACM Conference on Information and Knowledge Management (CIKM), in Glasgow, UK. The conference had a number of good papers in a wide range of topics, including information retrieval, social network analysis and machine learning. The conference also had two keynote speakers that talked about, respectively, improving user interfaces for better management of information, and addressing the challenges that persist in search, storage and analysis of biomedicine data.

I attended the first keynote, where David Karger, a professor from MIT, introduced some of the research outcomes of projects from the Haystack group.  Results showed better user experiences in terms of visualization, organization and sharing of information., the first project presented, is micro-note tool that can be added to Firefox. It focuses on the simplicity of adding simple, free-text notes right inside the browser, where users have the context at hand. In interviews, users highlighted ease/speed of use (35%), simplicity (20%), and “direct replacement for paper post-it” (20%) as the main reasons for using the tool. Usage patterns were also investigated, categorizing users in four main groups according to how they kept their notes.

Other tools, such as nb, were also concerned with keeping the context at hand. nb is a collaborative annotation tool integrated to a PDF reader. Notes can make references to contents of the PDF file and shared among other users of a class or group discussion. Similarly, FeedMe allows users to share in context a web page they are looking at. It also has a personalization component that recommends other users that are more likely to be interested in that information. Other projects, such as Datapress, were also presented with results from experiments. These show better user experiences when the tools kept the context of the user work and provided a personalized experience. More information can be obtained and the slides of the keynote can be downloaded from the official web site.

From the sessions, many good papers were presented. I will highlight three here. “A Probabilistic Method for Inferring Preferences from Clicks” utilizes interleaved comparison methods in click data from links in documents to learn ranking functions online. The Best Paper Award winner “Intent-Aware Query Similarity” by Guo, Cheng, Xu and Zhu besides proposing an interesting new method of computing query similarity, has a comprehensive review of similarity measures. Another good paper from Microsoft Research, “Semi-supervised Learning to Rank with Preference Regularization”, proposes a semi-supervised learning to rank algorithm. It builds on the assumption that similar items have similar preferences in terms of ranking.

Finally, on the last day of the conference, I attended the Second International Workshop on Web Science and Information Exchange in the Medical Web (MedEX 2011). Presentations focused on works investigating different questions related to web science in medicine and healthcare. In the workshop I presented the paper “Applicability of Recommender Systems to Medical Surveillance Systems” containing preliminary results of applying recommendation techniques to users of health surveillance systems. More details of the workshop and the slides of the presentation can be obtained from the official web site.