Situations, a General Framework for Studying Information Retrieval

This paper presents a framework for the theoretical comparison of information retrieval models based on how the models decide aboutness. The framework is based on concepts emerging from the eld of situation theory. So called infons and profons represent elementary information carriers which can be manipulated by union and fusion operators. These operators allow relationships between information carriers to be established. Sets of infons form so called situations which are used to model the information born by objects such as documents. An arbitrary information retrieval model can be mapped down into the framework. Special functions are de ned for this purpose depending on the model at hand. An important aspect is the inference mechanism which is mapped to inference between situations. Two examples are given based on the Boolean retrieval and coordination level matching models. The framework allows the comparison of retrieval models at an abstract level. Starting from an axiomatization of aboutness, retrieval models can be compared according to which axioms they are governed by. This approach is highlighted by the theoretical comparison of Boolean retrieval with coordinate level matching. This work was partly performed while employed at the Utrecht University.