PhD Research Seminar: Data Envelopment Analysis, Ontology-Mediated Queries

First talk: Data Envelopment Analysis Methods under Interval Estimates and Their Applications    
Speaker: Sergey Demin, second-year PhD student, Faculty of Economic Sciences

Data envelopment analysis (DEA) is one of the most popular techniques for the efficiency assessment of different objects with several inputs and outputs. Nowadays, it is widely used in various spheres, such as efficiency assessment of banks, hospitals, and universities.
However, the majority of DEA models can be applied only in the case of precise data, while real-world data usually contains some uncertainty. Therefore, we propose new DEA modifications based on interval parameter estimations. In addition, we will discuss properties and applications of the proposed methods.


Second talk: Checking the Data Complexity of Ontology-Mediated Queries      
Speaker: Olga Gerasimova, second-year PhD student, Faculty of Computer Science

In ontology-mediated query (OMQ) answering, a typical reasoning problem is to check whether a Boolean query q holds in every model of an ontology T and a data instance D.  In our research, we are concerned with a very special case of the problem: classify the OMQs with the fixed ontology Cov = {A → F or T} with concept names (unary predicates) A, F, and T and arbitrary conjunctive queries with unary and binary predicates according to their data complexity. It turns out that our single covering axiom  gives rise to surprisingly non-trivial and diverse classes of OMQs. To illustrate, we discuss a few examples. Also, we consider our attempts to understand the connection of our task to constraint satisfaction problems (CSPs).