SIGIR'17 Workshop on Axiomatic Thinking for Information Retrieval and Related Tasks (ATIR)


Introduction

This is the first workshop on the emerging interdisciplinary research area of applying axiomatic thinking to information retrieval (IR) and related tasks. The workshop aims to help foster collaboration of researchers working on different perspectives of axiomatic thinking and encourage discussion and research on general methodological issues related to applying axiomatic thinking to IR and re lated tasks. Please find more information about the motivation and theme of the workshop here.

Program (August 11, 2017)

  • 9:00-9:10: Opening remarks
  • 9:10-10:00: Keynote talk: "Inference in Axiomatic Approaches to IR", by Jian-Yun Nie
  • 10:00-10:20: Research Presentation 1: "On the Equivalence of Generative and Discriminative Formulations of the Sequential Dependence Model", by Laura Dietz and John Foley. (PDF)
  • 10:20-10:50: Coffee Break
  • 10:50-11:10: Research Presentation 2: "An Axiomatic Account of Similarity", by Enrique Amig√≥, Julio Gonzalo, Fernando Giner and Felisa Verdejo. (PDF, Poof)
  • 11:10-11:25: Research Presentation 3: "QM and IR: Another Perspective", by Paul Kantor. (PDF)
  • 11:25-11:40: Research Presentation 4: "On the Optimal Non-Personalized Recommendation: From the PRP to the Discovery False Negative Principle", by Rcio Canamares and Pablo Castells. (PDF)
  • 11:40-12:18: Panel Discussions: "Past, Present, and Fugure of Axiomatic Thinking". Moderated by ChengXiang Zhai.
  • 12:18-12:20: Closing/Wrap up
  • 12:20: Lunch break starts. The lunch will be provided at Hana at 4F of Keio Plaza Hotel.

Organizers

Keynote Talk

  • Speaker: Jian-Yun Nie, University of Montreal
  • Title: Inference in axiomatic approaches to IR
  • Abstract: The studies on axiomatic analysis in IR have found a set of intuitive constraints that a ranking function should satisfy. Up to now, the constraints are mainly related to term distributions in the document, the collection and the pseudo-feedback documents. They provide a good explanation of the success and failure of the traditional models. However, few studies have coped with the problem of inference and reasoning in IR, which is an important aspect in the current and future IR. Indeed, when we judge the relevance of a document, we not only look at the distribution of query terms in it, but also consider how the other terms are semantically related to the query. In this talk, we argue that this inference aspect should be covered in the axiomatic analysis. We will examine a few possible avenues to do it.

Panelists

  • Ben Carterette, University of Delaware
  • Pablo Castells, Universidad Autonoma de Madrid
  • Kevyn Colins-Thompson, University of Michigan
  • Paul Kantor, Rutgers University
  • Stephan E. Robertson, City University of London (Professor Emeritus)

Program Committee

  • Pablo Castells, Universidad Autonoma de Madrid
  • Fabio Crestani, University of Lugano
  • Ronan Cummins, University of Cambridge
  • Norbert Fuhr, University of Duisburg-Essen
  • Julio Gonzalo, UNED
  • Yuanhua Lv, Microsoft Research
  • Fabrizio Sebastiani, Qatar Computing Research Institute
  • Azadeh Shakery, University of Tehran

References