Inanç Seylan

4Machine learning papers tracked by Lacuna
Is this you?

Research Overview

Inanç Seylan focuses on knowledge representation, description logics, and the computational complexity of ontology-mediated querying. Their work investigates the theoretical properties of expressive logics, particularly regarding definability and the integration of closed-world assumptions into open-world frameworks. The most recent research addresses query rewriting algorithms and complexity dichotomies in ontology-based data access.

Research Trajectory

Seylan's early work in the corpus centers on the fundamental theoretical properties of description logics. As the senior author on Beth Definability in Expressive Description Logics (2011), Seylan investigated whether implicit definitions of concepts in a given signature guarantee the existence of explicit definitions within the same logic. This work provided a complete classification of Beth definability for extensions of the basic description logic ALC, considering both arbitrary and finite structures. This research contributes to the broader area of Finite Model Reasoning in Expressive Description Logics and touches upon themes related to Characterizing Logic Expressiveness via Bisimulation Invariance, establishing foundational limits and optimization strategies for reasoning systems.

By 2013, Seylan's focus expanded to the challenges of ontology-based data access (OBDA), particularly the tension between different logical assumptions. A major contribution during this period was the co-authored paper Ontology-Based Data Access with Closed Predicates is Inherently Intractable(Sometimes). This work addressed the research direction of Reconciling Open and Closed World Assumptions in Logic. Seylan and collaborators analyzed the impact of adopting the closed-world assumption for specific predicates within DL-Lite and EL ontologies. They demonstrated that answering conjunctive queries with closed predicates often leads to intractability, yielding complexity dichotomies between AC0, PTIME, and CONP.

In 2015, Seylan continued to explore the computational boundaries of ontology-mediated querying. As a second author on Ontology-Mediated Queries with Closed Predicates, Seylan further investigated the data complexity of queries where selected predicates are closed. This research tightly linked the complexity of closing concept names to surjective constraint satisfaction problems (CSPs) and proved that classifying the complexity when role names are closed is equivalent to classifying all problems in CONP. This work is a key example of the theme of Complexity Dichotomies in Ontology Mediated Querying.

Concurrently, Seylan participated as a co-author in developing practical algorithms for the area of First-Order Rewritability for Scalable Ontological Reasoning. The paper Efficient Query Rewriting in the Description Logic EL and Beyond proposed a novel, implementation-tailored algorithm for computing first-order rewritings of concept queries under ELHdr-TBoxes. The algorithm demonstrated that real-world ontologies often admit succinct non-recursive datalog rewritings even when theoretical guarantees are absent.

Current Directions

The most recent papers in the evidence, dating to 2015, indicate a focus on the intersection of theoretical complexity and practical query rewriting in ontology-based data access. A primary open problem addressed in this recent work is managing the computational cost of closed predicates in ontology-mediated queries, particularly mapping these challenges to constraint satisfaction problems to establish formal complexity dichotomies. Additionally, the work explores the development of efficient first-order rewriting algorithms that can handle real-world description logic ontologies, bridging the gap between theoretical intractability and practical performance in large-scale data management.

Profile updated

See something wrong? Report a mistake