# Structure of File Contents

These samples are related to the paper [1]. Refer to this paper for details on how the samples have
been used.

The sample file named `*_inst<i>_round<n>` contains the samples generated from instance `i` by
running the approach of Varga et al. [2] for `n` rounds. We publish sample files for instances
with generated availabilities (starting with `generated`) and for instances with availabilities
based on the Dutch time-use-survey (starting with `time-use-survey`). A file is separated into one
section per user, each section containing one test sample, which can also be used as training sample
by ignoring the fields `Ipred` and `Ipredhat`. Each section contains five fields, encoded with the
specifier followed by space separated values.

    - `Iprop`, `Iacc`, `Irej`: Proposed, accepted and rejected time intervals. An interval starting
       in timestep `tstart` and ending in timestep `tend` is represented by `tstart:tend`.
    - `Ipred`: Potential queries (type: time interval), to be used for testing a model.
    - `Ipredhat`: Whether the potential queries are accepted (value `true`) or rejected (value
      `false`). Values are in the same order as the queries in `Ipred`.

[1] Varga, J., Raidl, G.R., Rodemann, T.: Learning to Predict User Replies in
Interactive Job Scheduling. [submitted to CPAIOR 2025]

[2] Varga, J., Raidl, G.R., Rönnberg, E., Rodemann, T.: Scheduling jobs using queries
to interactively learn human availability times. Computers & Operations Research
167, 106648 (2024). https://doi.org/https://doi.org/10.1016/j.cor.2024.106648