Reproduce “Beyond States: Investigating the Effects of Context on User Modeling with Feature-Conditioned Markov Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jana Isabelle Friese, Andreas Konstantin Kruff, Timo Breuer, Philipp Schaer, Norbert Fuhr
The authors extend state-based search-user simulators so next-action probabilities depend on interaction state and contextual features, including rank, time, clicks, and document signals. They evaluate the models on three lab datasets, reusing observed queries and result lists. On TREC Session 2014, positional conditioning raises document-level click recall from 0.31 to 0.38; results vary by setting and metric.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.