The seminar is devoted to the theory and practice of data management and knowledge representation. We are interested in challenges related to the processing of data, queries, and metadata (schemas, constraints, dependencies, ontologies), ranging from designing and analyzing abstract formalisms all the way to database systems architecture and distributed processing of big data. We like our data in all flavors: not only relational, but also semistructured (XML, JSON), graph (RDF, LPG), object, text, temporal, stream, GIS, and others.
The problems tackled can be theoretical, requiring tools from algorithmics, combinatorics, logic (e.g. finite model theory), and automata theory, as well as very practical, in the spirit of systems and software engineering. MSc theses written within our seminar may study decidability and complexity of abstract problems, design algorithms and heuristics, implement and experiment with existing theoretical solutions, or analyze, compare and extend existing systems.
We meet and discuss with experts in other disciplines, who sometimes supply ideas for MSc theses. We have cooperated or are currently cooperating with astronomers, chemists, and geographers. We are also open for other areas where databases can be applied.
Seminar presentations are usually based on recent papers presented at leading international conferences devoted to data management and knowledge representation, such as VLDB, PODS, SIGMOD, or KR.
Selected topics:
* Data models, semantics, query languages
* Data provenance
* Databases for emerging hardware
* Distributed and parallel databases
* Graph data management, RDF, social networks, Semantic Web
* Knowledge discovery, clustering, data mining
* Machine learning for data management and vice versa
* Model theory, logics, algebras, computational complexity
* Ontology-based data access, data integration and exchange, metadata management
* Ontology formalisms and models, description logics
* Privacy, security, ethics
* Query processing and optimization
* Scientific databases
* Semi-structured data
* Small data, end-user programming
* Storage, indexing, and physical database design
* Streams, sensor networks, complex event processing
* Transaction processing
* Uncertainty, incompleteness, and inconsistency in data management
Organizers
- dr hab. Filip Murlak, prof. ucz.
- dr Jacek Sroka
- prof. dr hab. Krzysztof Stencel
- prof. dr hab. Jerzy Tyszkiewicz
Information
Tuesdays, 10:15 a.m. , room: 4060Home page
https://sites.google.com/view/sembdmimuw?pli=1&authuser=1Research fields
List of talks
-
May 27, 2025, 10:15 a.m.
Damian Werpachowski (MIMUW)
Implementation of UDP network stack for Java using ef_vi (Implementation of UDP network stack for Java using ef_vi)
-
May 20, 2025, 10:15 a.m.
Bartosz Ruszewski (MIMUW)
Evaluation and Enumeration of Regular Simple Path and Trail Queries (Evaluation and Enumeration of Regular Simple Path and Trail Queries)
-
May 13, 2025, 10:15 a.m.
Piotr Ulanowski (MIMUW)
Ewaluacja zapytań ścieżkowych w językach GQL i SQL/PGQ z wykorzystaniem różnych algorytmów przeszukiwania grafów
-
May 6, 2025, 10:15 a.m.
Michał Molas (MIMUW)
ALEX: Cardinality Estimation of LIKE Predicate Queries using Deep Learning (ALEX: Cardinality Estimation of LIKE Predicate Queries using Deep Learning)
-
April 29, 2025, 10:45 a.m.
Zuzanna Surowiec (MIMUW)
Optimizing Nested Recursive Queries (Optimizing Nested Recursive Queries)
-
April 29, 2025, 10:15 a.m.
Marta Jadwiga Burzańska (UMK)
Heuristic algorithm for periodic patterns discovery in a database workload reconstruction (Heuristic algorithm for periodic patterns discovery in a database workload reconstruction)
Information about the existence of periodic patterns in a database workload can play a big part in the process of database tuning. However, full analysis of audit trails can be cumbersome and time-consuming. This talk …
-
April 15, 2025, 10:15 a.m.
Jakub Kłos (MIMUW)
Differentially Private Data Release over Multiple Tables (Differentially Private Data Release over Multiple Tables)
-
April 8, 2025, 10:15 a.m.
Katarzyna Mielnik (MIMUW)
Efficiently Processing Joins and Grouped Aggregations on GPUs (Efficiently Processing Joins and Grouped Aggregations on GPUs)
-
March 25, 2025, 10:15 a.m.
Marcin Mordecki (MIMUW)
Analiza wpływu wykorzystania instrukcji SIMD na wydajność przetwarzania
W pierwszym semestrze przyjrzeliśmy się architekturze SIMD oraz potencjalnym zyskom i pułapkom, jakie wiążą się z wektoryzacją kodu za pomocą instrukcji AVX. Kontynuując ten temat, tym razem zbadamy, co dokładnie daje wykorzystanie najnowszego zestawu tychże …
-
March 18, 2025, 10:15 a.m.
Michał J. Gajda (Well. co)
Zamienianie tabel w strumienie zdarzeń przyrostowych i odwrotnie
Ze względu na wydajność, duże bazy danych często utrzymuje się w ten sposób, że dzielimy je na strumienie zdarzeń i "zmaterializowane" tabele lub perspektywy. W zależności od zastosowania chcielibyśmy przetwarzać przyrostowy strumień zdarzeń albo tabelę …
-
March 4, 2025, 10:15 a.m.
Krzysztof Żyndul (MIMUW)
ALEX: An Updatable Adaptive Learned Index (ALEX: An Updatable Adaptive Learned Index)
-
Feb. 25, 2025, 10:15 a.m.
Alexandra Rogova (MIMUW)
Dangers of List Processing in Querying Property Graphs (Dangers of List Processing in Querying Property Graphs)
The focus of graph databases is graph-like data, i.e. data that represents heavily-linked information where the topology is an important aspect. The workhorse of graph query languages is pattern matching. The result of pattern matching …
-
Jan. 21, 2025, 10:15 a.m.
Damian Werpachowski (MIMUW)
Implementation of UDP network stack for Java using ef_vi (Implementation of UDP network stack for Java using ef_vi)
-
Jan. 14, 2025, 10:15 a.m.
Michał Molas (MIMUW)
LadderFilter: Filtrowanie rzadkich elementów przy niewielkim zużyciu pamięci i czasu (LadderFilter: Filtering Infrequent Items with Small Memory and Time Overhead)
-
Jan. 7, 2025, 10:15 a.m.
Katarzyna Mielnik (MIMUW)
Lemo: A Cache-Enhanced Learned Optimizer for Concurrent Queries (Lemo: A Cache-Enhanced Learned Optimizer for Concurrent Queries)
Realizacja wielu zapytań w krótkim czasie ma szerokie zastosowanie praktyczne. Aby jednak osiągnąć wysoką wydajność, kluczowe jest zminimalizowanie powtarzających się obliczeń oraz opracowanie efektywnego planu wykonania współbieżnych zapytań. W metodzie Lemo zastosowano wytrenowaną sieć, która …