We consider the problem of answering conjunctive queries with aggregation on database instances that may violate primary key constraints. In SQL, these queries follow the SELECT-FROM-WHERE-GROUP BY format, where the WHERE clause involves a conjunction of equalities, and the SELECT clause can incorporate aggregate operators like MAX, MIN, SUM, AVG, or COUNT. Repairs of a database instance are defined as inclusion-maximal subsets that satisfy all primary keys. The range-consistent answer to a numerical query over an inconsistent database is a pair [glb, lub], where glb and lub are, respectively, the smallest and the greatest results returned by the query over all possible repairs. While previous work has focused on the computation of the glb, the current paper studies the computation of the lub for a numerical domain of non-negative rational numbers. We introduce the notion of kappa-acyclicity for self-join-free conjunctive queries. We show that if the body of a SUM-query is kappa-acyclic, then the lub can be computed through a rewriting in first-order aggregate logic. Moreover, we show that this result extends to all aggregate operators that are monotone and associative. Importantly, we also prove the inverse: if the body of a SUM-query is not kappa-acyclic, then the lub cannot be computed in first-order aggregate logic.
Previous research on join sampling has focused on joins without selection conditions, even though such conditions are prevalent in everyday queries in database systems. Motivated by this, we undertake a systematic investigation on the complexity of sampling from the result of an acyclic join under equality conditions given only at runtime. When conditions are conjunctive, the goal is to understand when it is possible to precompute a feasible structure that uses (O) over tilde (IN) space and supports sampling in (O) over tilde (1) time, where IN is the input size. We present a dichotomy to characterize (subject to a widely-accepted conjecture) the existence of such structures based on the conditions supplied and, in every feasible scenario, give an optimal structure of O(IN) space and O(1) sample time. We then extend our investigation to conditions expressed in disjunctive normal form, where the core challenge reduces to the fundamental set union sampling problem. We overcome the challenge with an optimal algorithm and utilize it to develop optimal sampling structures. Our findings also lead to new results on the closely-related random enumeration problem.
Important learning objectives in database education are to learn how to design database schemas and how to write queries to access data stored in a database adhering to a schema. In this article we report on results from [7](1) where this learning objective is addressed for the related educational task of modelling with logical formalisms. Two key steps in logical modelling are to (a) choose a suitable vocabulary, that is, e.g., which first-order symbols to use and with which intended meaning, and then to (b) construct actual formal descriptions, i.e. first-order formulas over the chosen vocabulary. While (b) is addressed by several educational support systems for formal foundations of computer science, (a) is so far not addressed at all - likely because it involves specifying the intended meaning of symbols in natural language. We propose a conceptual framework for educational tasks where students choose a vocabulary and implement it for tasks for designing propositional and first-order vocabularies within the Iltis educational system.
The view-based query determinacy problem asks, given a view V and a query Q whether the information contained in V always suffices to answer the query Q. It is a notoriously hard problem that is known to be undecidable even in very restricted settings. Here, we study it in the context of graph databases and regular path queries and views evaluated under several "walk-based semantics", that is, semantics in which queries and views return the matched walks in full as opposed to the classical "endpoint semantics". Our main finding is that view-based query determinacy is decidable for regular path queries under both trail and shortest walk semantics.
Aggregate queries often require computing large intermediate joins despite producing only small outputs. We identify broad classes of acyclic aggregate queries that can be evaluated without materialising any join results, using a bottom-up, semi-join-based propagation of cardinalities and partial aggregates. An implementation in Spark SQL shows that this approach is widely applicable and yields substantial performance gains on standard benchmarks.
We study the worst-case communication complexity of the join query evaluation problem over large-scale data in distributed shared-nothing systems under the MPC model. We focus on multi-round MPC algorithms that run in constant number of rounds. The problem is well-understood for a few classes of queries, mainly the class of acyclic queries and the class of graph-like queries. For queries not belonging to either class, the complexity picture is much less clear. We study the class of degree-two queries and fragments thereof. In this paper, we tighten the gap between the upper and lower bounds for the studied classes and establish worst-case optimality for some fragments of the considered classes. We also debunk a well-believed conjecture about which query-related quantity, in the worst-case, optimally captures the communication complexity of the studied problem.
The performance of transactional database systems is typically evaluated by measuring the amount of transactions they can commit to the database per second. However, fairly measuring this for the same workload on different systems is not trivial. It is therefore relevant to formalize schedule efficiency, investigate the space of all possible efficient schedules, and identify whether there is any room for improvement. Prior transaction theory largely centers on decision problems relating to safety, such as the serializability, robustness, and allocation problems. Most pertinently, these problems take already scheduled transactions as input, and do not directly consider the efficiency of those schedules. In this work, we define schedules as assignments of operations on objects to discrete points in time. This allows us to quantify efficiency as the elapsed duration between the schedule's beginning and end, more commonly known as the makespan in the scheduling literature. We establish that, given some set of transactions and a desired makespan, it is NP-complete to decide if there exists a conflict serializable schedule which is bounded by that makespan. We additionally provide an instance optimal algorithm for scheduling transaction sets with a single contention point, that is, exactly one object may appear in conflicting operations. Lastly, we give worst-case optimal bounds on the makespan, meaning that schedules can never exceed this bound, and for the worst transaction sets, the bound is optimal.
The idea of using an ontology to enrich user queries with domain knowledge has attracted considerable attention from the database and KR communities during the last fifteen years or so. The ontology and the user query can be conveniently seen as two components of one composite query, called ontology-mediated query (omq), while an omq language (OL, QL) collects all such omqs where the ontology is expressed using the ontology language OL and the user query comes from the query language QL. The evaluation problem for rule-based omq languages of the form (OL, CQ), where OL is a rule-based ontology language, i.e., it collects ontologies modelled using tuple-generating dependencies (a.k.a. existential rules), and CQ is the language of conjunctive queries, has been extensively studied in the literature. In particular, the notion of first-order rewritability of such languages, i.e., the property of being able to rewrite every omq from the language in question to an equivalent first-order query, has been studied in depth. This research effort led an algorithmic characterization of when a rule-based omq language (OL, CQ) is first-order rewritable. More precisely, there is a uniform algorithm Rewrite such that, for every rule-based ontology language OL, the omq language (OL, CQ) is first-order rewritable iff for every omq O from (OL, CQ), the algorithm Rewrite on input O terminates and constructs a first-order rewriting of O. The question that we are interested in is whether the above algorithmic characterization can be extended to rule-based omq languages of the form (OL, CQ(sic)), where CQ (sic) is the language of conjunctive queries with the useful feature of negation. The goal of this work is to initiate effort towards the settlement of the above highly non-trivial question. To this end, we provide a new algorithm, which is a non-trivial extension of the algorithm Rewrite for positive omqs, and show the following: under the Skolem semantics, a well-established approach for defining the answer to a rule-based omq when the user query can use negation, the proposed algorithm is a first-order rewriter for (OL, CQ (sic)), where OL is the language of linear or acyclic tuple-generating dependencies, two central rule-based ontology languages that ensure first-order rewritability for positive omqs. We strongly believe that the new algorithm can serve as a good starting point towards the full settlement of our main question.
Query containment is a fundamental database problem which has been extensively studied for conjunctive queries (CQs). The most famous result is arguably the Homomorphism Theorem: a CQ q(1) is contained in a CQ q(2) iff there is a homomorphism from q(2) to q(1). However, when extending conjunctive queries with safe base negation (CQ (sic)), this test becomes incomplete, hence, requiring significantly more expensive procedures due to its inherently harder complexity (Pi(P)(2)-hard). In this paper, we define and study the classes CQ(HT)(1(sic)) and CQ(HT)((sic)): the classes of conjunctive queries extended with one or several safe negated atoms that satisfy the Homomorphism Theorem, and hence, whose containment check is in NP. To characterise them, we define what we call the dependency-version of a query, which is a dependency that, intuitively, models the databases in which the query is false. It turns out that, when the query q contains one (several) negated atom(s), the query satisfies the Homomorphism Theorem iff its tgd(ded)-version is uniformly one-bounded. We also show that CQ(HT)((sic)) membership is EXPTIME-hard, but its complexity reduces to Pi(P)(2) in the CQ(HT)(1(sic)) case, and to NP when bounding the number of positive atoms that can unify with the negated one.
Modern property graph database query languages such as Cypher, PGQL, GSQL, and the standard GQL draw inspiration from the formalism of regular path queries (RPQs). In order to output walks explicitly, they depart from the classical and well-studied homomorphism semantics. However, it then becomes difficult to present results to users because RPQs may match infinitely many walks. The aforementioned languages use ad-hoc criteria to select a finite subset of those matches. For instance, Cypher uses trail semantics, discarding walks with repeated edges; PGQL and GSQL use shortest walk semantics, retaining only the walks of minimal length among all matched walks; and GQL allows users to choose from several semantics. Even though there is academic research on these semantics, it focuses almost exclusively on evaluation efficiency. In an attempt to better understand, choose and design RPQ semantics, we present a framework to categorize and compare them according to other criteria. We formalize several possible properties, pertaining to the study of RPQ semantics seen as mathematical functions mapping a database and a query to a finite set of walks. We show that some properties are mutually exclusive, or cannot be met. We also give several new RPQ semantics as examples. Some of them may provide ideas for the design of new semantics for future graph database query languages.
The rise of generative AI as a primary information source presents a paradigm shift from traditional web search. This paper presents a large-scale empirical study quantifying the fundamental differences between the results returned by Google Search and leading generative AI services. We analyze multiple dimensions, demonstrating that AI-generated answers and web search results diverge significantly in their consulted source domains, the typology of these domains (e.g., earned media vs. owned, social), query intent and the freshness of the information provided. We then investigate the role of LLM pre-training as a key factor shaping these differences, analyzing how this intrinsic knowledge base interacts with and influences real-time web search when enabled. Our findings reveal the distinct mechanics of these two information ecosystems, leading to critical observations on the emergent field of Answer Engine Optimization (AEO) and its contrast with traditional Search Engine Optimization (SEO).
Learned index structures aim to accelerate queries by training machine learning models to approximate the rank function associated with a database attribute. While effective in practice, their theoretical limitations are not fully understood. We present a framework for proving lower bounds on query time for learned indexes, expressed in terms of their space overhead and parameterized by the model class used for approximation. Our formulation captures a broad family of one-dimensional learned indexes, including most existing designs, as piecewise model-based predictors. We solve the problem of lower bounding query time in two steps: first, we use probabilistic tools to control the effect of sampling when the database attribute is drawn from a probability distribution. Then, we analyze the approximation-theoretic problem of how to optimally represent a cumulative distribution function with approximators from a given model class. Within this framework, we derive lower bounds under a range of modeling and distributional assumptions, paying particular attention to the case of piecewise linear and piecewise constant model classes, which are common in practical implementations. Our analysis shows how tools from approximation theory, such as quantization and Kolmogorov widths, can be leveraged to formalize the space-time trade-offs inherent to learned index structures. The resulting bounds illuminate core limitations of these methods.
We investigate the problem of identifying database repairs for missing tuples in query answers. We show that when the query is part of the input - the combined complexity setting - determining whether or not a repair exists is polynomial-time equivalent to the satisfiability problem for classes of queries admitting a weak form of projection and selection. We then identify the sub-classes of unions of conjunctive queries with negated atoms, defined by the relational algebra operations permitted to appear in the query, for which the minimal repair problem can be solved in polynomial time. In contrast, we show that the problem is NP-hard, as well as set cover-hard to approximate via strict reductions, whenever both projection and join are permitted in the input query. Additionally, we show that finding the size of a minimal repair for unions of conjunctive queries (with negated atoms permitted) is OptP[log(n)]-complete, while computing a minimal repair is possible with O(n(2)) queries to an NP oracle. With recursion permitted, the combined complexity of all of these variants increases significantly, with an EXP lower bound. However, from the data complexity perspective, we show that minimal repairs can be identified in polynomial time for all queries expressible as semi-positive datalog programs.
How important is the weight of a given column in determining the ranking of tuples in a table? To address such an explanation question about a ranking function, we investigate the computation of SHAP scores for column weights, adopting a recent framework by Grohe et al.[ICDT'24]. The exact definition of this score depends on three key components: (1) the ranking function in use, (2) an effect function that quantifies the impact of using alternative weights on the ranking, and (3) an underlying weight distribution. We analyze the computational complexity of different instantiations of this framework for a range of fundamental ranking and effect functions, focusing on probabilistically independent finite distributions for individual columns. For the ranking functions, we examine lexicographic orders and score-based orders defined by the summation, minimum, and maximum functions. For the effect functions, we consider global, top-k, and local perspectives: global measures quantify the divergence between the perturbed and original rankings, top-k measures inspect the change in the set of top-k answers, and local measures capture the impact on an individual tuple of interest. Although all cases admit an additive fully polynomial-time randomized approximation scheme (FPRAS), we establish the complexity of exact computation, identifying which cases are solvable in polynomial time and which are #P-hard. We further show that all complexity results, lower bounds and upper bounds, extend to a related task of computing the Shapley value of whole columns (regardless of their weight).
Direct access asks for the retrieval of query answers by their ranked position, given a query and a desired order. While the time complexity of data structures supporting such accesses has been studied in depth, and efficient algorithms for many queries and common orders are known, their practical performance has received little attention. We provide an implementation covering a wide range of queries and orders; it allows us to investigate intriguing practical aspects, including the comparative performance of database systems and the relationship between direct access and its single-access counterpart.
