Learning to Quantify (LQ - also known as “quantification“, or “supervised prevalence estimation“, or “class prior estimation“), is the task of training class prevalence estimators via supervised learning. In other words, the task of these trained models is to estimate, given an unlabelled sample of data items and a set of classes, the prevalence (i.e., relative frequency) of each such class in the sample.

LQ is interesting in all applications of classification in which the final goal is not determining which class (or classes) individual unlabelled data items belong to, but estimating the percentages of data items that belong to the classes of interest, i.e., estimating the distribution of the unlabelled data items across the classes. Example disciplines whose interest in labelling data items is at the aggregate level (rather than at the individual level), are the social sciences, political science, market research, ecological modelling, and epidemiology.

While LQ may in principle be solved by classifying each data item in the sample and counting how many such items have been labelled with a certain class, it has been shown that this “classify and count” method yields suboptimal quantification accuracy. As a result, quantification is now no longer considered a mere byproduct of classification, and has evolved as a task of its own.

The goal of this workshop is to bring together all researchers interested in methods, algorithms, evaluation measures, and methodologies for LQ, as well as practitioners interested in their practical application to managing large quantities of data.

We seek original, unpublished papers on topics of relevance to LQ; also papers currently under submission at other workshops / conferences / journals are welcome, provided this double submission does not violate the rules of these workshops / conferences / journals.

We seek papers on any of the following topics, which will form the main themes of the workshop:

  • Binary, multiclass, and ordinal LQ
  • Supervised algorithms for LQ
  • Semi-supervised / transductive LQ
  • Deep learning for LQ
  • Representation learning for LQ
  • LQ and dataset shift
  • Evaluation measures for LQ
  • Experimental protocols for the evaluation of LQ
  • Quantification of streaming data
  • Quantifying text by topic and quantifying text by sentiment
  • Novel applications of LQ

and other topics of relevance to LQ.

Important dates

all 23:59 AoE

  • Paper submission deadline: August 31, 2021
  • A/R notification deadline: September 21, 2021
  • Final copy submission deadline: October 10, 2021

Submission

Submit your paper to LQ 2021 using the EasyChair system.

Papers should be formatted according to the same format as for the main CIKM 2021 conference (see https://www.cikm2021.org/cfp/full-papers), and should be up to 9 pages + references in length; however, this is just the upper bound, and contributions of any length up to this bound will be considered.

The workshop will be held entirely online. At least one author of each accepted paper must register to present “live” the work as scheduled in the workshop program. The proceedings of the workshop will be published, together with the proceedings of other CIKM 2021 workshops, in the CEUR series of free open-access proceedings for scientific conferences and workshops (http://ceur-ws.org/). However, for each accepted paper, it will be left at the discretion of the authors to decide whether to contribute their paper or not to the proceedings.