Information Retrieval Journal

Special Issue on Learning to Rank for Information Retrieval

-------------------------------------------------------

*******Submission deadline: March 27, 2009********

** Introduction **

Learning to rank has emerged as an active and growing area of research both in information retrieval (IR) and machine learning (ML). The goal of learning to rank is to automatically learn a ranking model from training data, such that the model can sort objects (e.g., documents) according to their degrees of relevance, preference, or importance as defined in a specific application. Many IR problems are by nature ranking problems, and many IR technologies can be potentially enhanced by using learning to rank techniques.


In recent years, many learning to rank papers have been published at major conferences and journals in the IR and ML communities. At this stage, several questions regarding this research direction naturally arise. For example,

- To what respect are existing learning to rank algorithms similar and in which aspects do they differ? What are the strength and weakness of them?

- What kind of loss function is the most suitable one for learning to rank?

- How to create effective training data for learning to rank?

- What is the role of features in the task of learning to rank? What is the relationship between features and loss function?

- How to solve complex IR tasks (such as search result diversification, topic distillation, etc.) using learning to rank technologies?

- What is unique for ranking as compared to other machine learning problems (regression, classification, etc.)? Is ranking a new machine learning problem, or can be simply reduced to existing machine learning problems?

- What are the future research directions for learning to rank?


The Information Retrieval Journal is pleased to announce a special issue on learning to rank for IR. High quality papers trying to (but not limited to) answer the above questions with theoretical and/or experimental orientation are sought.


** Indicative list of topics **


The special issue welcomes high-quality manuscripts on the development, evaluation, and theoretical analysis of learning to rank for IR. Submissions to the special issue should not be under consideration in any other journal or conference and will be evaluated according to the Information Retrieval Journal reviewing criteria and appropriateness to the special issue. If the submission is a revision of a previous conference paper, the revision must contain significant amplification or clarification of the original material or there is some significant additional benefit to be gained. For more details, please refer to "manuscript submission" on the journal homepage (www.springer.com/computer/database+management+%26+information+retrieval/journal/10791) .


Topics of interest include, but are not limited, to:

- Models, features, and algorithms for learning to rank

- Theoretical analyses of learning to rank

- Data creation methods for learning to rank

- Empirical comparisons between learning to rank methods

- Comparisons between traditional approaches and learning approaches to ranking


Submissions should use the Information Retrieval Journal style templates available from the Journal's homepage (http://www.springer.com/computer/database+management+%26+information+retrieval/journal/10791) and should be submitted through the IR Journal online submission page (https://www.editorialmanager.com/inrt/) selecting the 'S.I.: Learning to Rank for Info. Ret.' article type.


** Important dates **

Submission of articles: March 27, 2009.

Notification to authors: June 12, 2009.

All questions regarding submissions should be directed to the special issue Guest Editors.


- Thorsten Joachims (tj@cs.cornell.edu, Cornell University),

- Hang Li (hangli@microsoft.com, Microsoft Research Asia),

- Tie-Yan Liu (tyliu@microsoft.com, Microsoft Research Asia),

- Chengxiang Zhai (czhai@cs.uiuc.edu, University of Illinois at Urbana-Champaign).