Search

This document is an introduction to two related formalisms to define Boolean functions: binary decision diagrams, and Boolean circuits. It presents these formalisms and several of their variants studied in the setting of knowledge compilation. Last, it explains how these formalisms can be connected to the notions of automata over words and trees., Comment: 26 pages

In this note, we present a conjecture on intersections of set families, and a rephrasing of the conjecture in terms of principal downsets of Boolean lattices. The conjecture informally states that, whenever we can express the measure of a union of sets in terms of the measure of some of their intersections using the inclusion-exclusion formula, then we can express the union as a set from these same intersections via the set operations of disjoint union and subset complement. We also present a partial result towards establishing the conjecture., Comment: 30 pages

Shapley values, originating in game theory and increasingly prominent in explainable AI, have been proposed to assess the contribution of facts in query answering over databases, along with other similar power indices such as Banzhaf values. In this work we adapt these Shapley-like scores to probabilistic settings, the objective being to compute their expected value. We show that the computations of expected Shapley values and of the expected values of Boolean functions are interreducible in polynomial time, thus obtaining the same tractability landscape. We investigate the specific tractable case where Boolean functions are represented as deterministic decomposable circuits, designing a polynomial-time algorithm for this setting. We present applications to probabilistic databases through database provenance, and an effective implementation of this algorithm within the ProvSQL system, which experimentally validates its feasibility over a standard benchmark., Comment: 27 pages, including 20 pages of maintext. This is the authors' version of the corresponding PODS'2024 article

Attribution scores can be applied in data management to quantify the contribution of individual items to conclusions from the data, as part of the explanation of what led to these conclusions. In Artificial Intelligence, Machine Learning, and Data Management, some of the common scores are deployments of the Shapley value, a formula for profit sharing in cooperative game theory. Since its invention in the 1950s, the Shapley value has been used for contribution measurement in many fields, from economics to law, with its latest researched applications in modern machine learning. Recent studies investigated the application of the Shapley value to database management. This article gives an overview of recent results on the computational complexity of the Shapley value for measuring the contribution of tuples to query answers and to the extent of inconsistency with respect to integrity constraints. More specifically, the article highlights lower and upper bounds on the complexity of calculating the Shapley value, either exactly or approximately, as well as solutions for realizing the calculation in practice., Comment: 12 pages, including references. This is the authors version of the corresponding SIGMOD Record article

We study the problem of enumerating the satisfying assignments for circuit classes from knowledge compilation, where assignments are ranked in a specific order. In particular, we show how this problem can be used to efficiently perform ranked enumeration of the answers to MSO queries over trees, with the order being given by a ranking function satisfying a subset-monotonicity property. Assuming that the number of variables is constant, we show that we can enumerate the satisfying assignments in ranked order for so-called multivalued circuits that are smooth, decomposable, and in negation normal form (smooth multivalued DNNF). There is no preprocessing and the enumeration delay is linear in the size of the circuit times the number of values, plus a logarithmic term in the number of assignments produced so far. If we further assume that the circuit is deterministic (smooth multivalued d-DNNF), we can achieve linear-time preprocessing in the circuit, and the delay only features the logarithmic term., Comment: 26 pages; this is the authors version of the corresponding ICDT'24 article

We study the task, for a given language $L$, of enumerating the (generally infinite) sequence of its words, without repetitions, while bounding the delay between two consecutive words. To allow for delay bounds that do not depend on the current word length, we assume a model where we produce each word by editing the preceding word with a small edit script, rather than writing out the word from scratch. In particular, this witnesses that the language is orderable, i.e., we can write its words as an infinite sequence such that the Levenshtein edit distance between any two consecutive words is bounded by a value that depends only on the language. For instance, $(a+b)^*$ is orderable (with a variant of the Gray code), but $a^* + b^*$ is not. We characterize which regular languages are enumerable in this sense, and show that this can be decided in PTIME in an input deterministic finite automaton (DFA) for the language. In fact, we show that, given a DFA $A$, we can compute in PTIME automata $A_1, \ldots, A_t$ such that $L(A)$ is partitioned as $L(A_1) \sqcup \ldots \sqcup L(A_t)$ and every $L(A_i)$ is orderable in this sense. Further, we show that the value of $t$ obtained is optimal, i.e., we cannot partition $L(A)$ into less than $t$ orderable languages. In the case where $L(A)$ is orderable (i.e., $t=1$), we show that the ordering can be produced by a bounded-delay algorithm: specifically, the algorithm runs in a suitable pointer machine model, and produces a sequence of bounded-length edit scripts to visit the words of $L(A)$ without repetitions, with bounded delay -- exponential in $|A|$ -- between each script. In fact, we show that we can achieve this while only allowing the edit operations push and pop at the beginning and end of the word, which implies that the word can in fact be maintained in a double-ended queue., Comment: This is the full versions with proofs of the STACS'23 article

We consider a weighted counting problem on matchings, denoted $\textrm{PrMatching}(\mathcal{G})$, on an arbitrary fixed graph family $\mathcal{G}$. The input consists of a graph $G\in \mathcal{G}$ and of rational probabilities of existence on every edge of $G$, assuming independence. The output is the probability of obtaining a matching of $G$ in the resulting distribution, i.e., a set of edges that are pairwise disjoint. It is known that, if $\mathcal{G}$ has bounded treewidth, then $\textrm{PrMatching}(\mathcal{G})$ can be solved in polynomial time. In this paper we show that, under some assumptions, bounded treewidth in fact characterizes the tractable graph families for this problem. More precisely, we show intractability for all graph families $\mathcal{G}$ satisfying the following treewidth-constructibility requirement: given an integer $k$ in unary, we can construct in polynomial time a graph $G \in \mathcal{G}$ with treewidth at least $k$. Our hardness result is then the following: for any treewidth-constructible graph family $\mathcal{G}$, the problem $\textrm{PrMatching}(\mathcal{G})$ is intractable. This generalizes known hardness results for weighted matching counting under some restrictions that do not bound treewidth, e.g., being planar, 3-regular, or bipartite; it also answers a question left open in Amarilli, Bourhis and Senellart (PODS'16). We also obtain a similar lower bound for the weighted counting of edge covers., Comment: This is the full version with proofs of the MFCS'22 article

The Shapley value is a game-theoretic notion for wealth distribution that is nowadays extensively used to explain complex data-intensive computation, for instance, in network analysis or machine learning. Recent theoretical works show that query evaluation over relational databases fits well in this explanation paradigm. Yet, these works fall short of providing practical solutions to the computational challenge inherent to the Shapley computation. We present in this paper two practically effective solutions for computing Shapley values in query answering. We start by establishing a tight theoretical connection to the extensively studied problem of query evaluation over probabilistic databases, which allows us to obtain a polynomial-time algorithm for the class of queries for which probability computation is tractable. We then propose a first practical solution for computing Shapley values that adopts tools from probabilistic query evaluation. In particular, we capture the dependence of query answers on input database facts using Boolean expressions (data provenance), and then transform it, via Knowledge Compilation, into a particular circuit form for which we devise an algorithm for computing the Shapley values. Our second practical solution is a faster yet inexact approach that transforms the provenance to a Conjunctive Normal Form and uses a heuristic to compute the Shapley values. Our experiments on TPC-H and IMDB demonstrate the practical effectiveness of our solutions.

In Machine Learning, the $\mathsf{SHAP}$-score is a version of the Shapley value that is used to explain the result of a learned model on a specific entity by assigning a score to every feature. While in general computing Shapley values is an intractable problem, we prove a strong positive result stating that the $\mathsf{SHAP}$-score can be computed in polynomial time over deterministic and decomposable Boolean circuits. Such circuits are studied in the field of Knowledge Compilation and generalize a wide range of Boolean circuits and binary decision diagrams classes, including binary decision trees and Ordered Binary Decision Diagrams (OBDDs). We also establish the computational limits of the SHAP-score by observing that computing it over a class of Boolean models is always polynomially as hard as the model counting problem for that class. This implies that both determinism and decomposability are essential properties for the circuits that we consider. It also implies that computing $\mathsf{SHAP}$-scores is intractable as well over the class of propositional formulas in DNF. Based on this negative result, we look for the existence of fully-polynomial randomized approximation schemes (FPRAS) for computing $\mathsf{SHAP}$-scores over such class. In contrast to the model counting problem for DNF formulas, which admits an FPRAS, we prove that no such FPRAS exists for the computation of $\mathsf{SHAP}$-scores. Surprisingly, this negative result holds even for the class of monotone formulas in DNF. These techniques can be further extended to prove another strong negative result: Under widely believed complexity assumptions, there is no polynomial-time algorithm that checks, given a monotone DNF formula $\varphi$ and features $x,y$, whether the $\mathsf{SHAP}$-score of $x$ in $\varphi$ is smaller than the $\mathsf{SHAP}$-score of $y$ in $\varphi$., Comment: Up to the formatting, this is the exact content of the paper in Journal of Machine Learning Research (JMLR)

We study the complexity of various fundamental counting problems that arise in the context of incomplete databases, i.e., relational databases that can contain unknown values in the form of labeled nulls. Specifically, we assume that the domains of these unknown values are finite and, for a Boolean query $q$, we consider the following two problems: given as input an incomplete database $D$, (a) return the number of completions of $D$ that satisfy $q$; or (b) return the number of valuations of the nulls of $D$ yielding a completion that satisfies $q$. We obtain dichotomies between \#P-hardness and polynomial-time computability for these problems when $q$ is a self-join-free conjunctive query, and study the impact on the complexity of the following two restrictions: (1) every null occurs at most once in $D$ (what is called Codd tables); and (2) the domain of each null is the same. Roughly speaking, we show that counting completions is much harder than counting valuations: for instance, while the latter is always in \#P, we prove that the former is not in \#P under some widely believed theoretical complexity assumption. Moreover, we find that both (1) and (2) can reduce the complexity of our problems. We also study the approximability of these problems and show that, while counting valuations always has a fully polynomial-time randomized approximation scheme (FPRAS), in most cases counting completions does not. Finally, we consider more expressive query languages and situate our problems with respect to known complexity classes., Comment: 51 pages, including 43 pages of main text. Extended version of arXiv:1912.11064. Up to the stylesheet, page/environment numbering, minor formatting, and publisher-induced changes, this is the exact content of the paper in ACM Transactions on Computational Logic

In spite of several claims stating that some models are more interpretable than others -- e.g., "linear models are more interpretable than deep neural networks" -- we still lack a principled notion of interpretability to formally compare among different classes of models. We make a step towards such a notion by studying whether folklore interpretability claims have a correlate in terms of computational complexity theory. We focus on local post-hoc explainability queries that, intuitively, attempt to answer why individual inputs are classified in a certain way by a given model. In a nutshell, we say that a class $\mathcal{C}_1$ of models is more interpretable than another class $\mathcal{C}_2$, if the computational complexity of answering post-hoc queries for models in $\mathcal{C}_2$ is higher than for those in $\mathcal{C}_1$. We prove that this notion provides a good theoretical counterpart to current beliefs on the interpretability of models; in particular, we show that under our definition and assuming standard complexity-theoretical assumptions (such as P$\neq$NP), both linear and tree-based models are strictly more interpretable than neural networks. Our complexity analysis, however, does not provide a clear-cut difference between linear and tree-based models, as we obtain different results depending on the particular post-hoc explanations considered. Finally, by applying a finer complexity analysis based on parameterized complexity, we are able to prove a theoretical result suggesting that shallow neural networks are more interpretable than deeper ones., Comment: 36 pages, including 9 pages of main text. This is the arXiv version of the NeurIPS'2020 paper. Except from minor differences that could be introduced by the publisher, the only difference should be the addition of the appendix, which contains all the proofs that do not appear in the main text

Scores based on Shapley values are widely used for providing explanations to classification results over machine learning models. A prime example of this is the influential SHAP-score, a version of the Shapley value that can help explain the result of a learned model on a specific entity by assigning a score to every feature. While in general computing Shapley values is a computationally intractable problem, it has recently been claimed that the SHAP-score can be computed in polynomial time over the class of decision trees. In this paper, we provide a proof of a stronger result over Boolean models: the SHAP-score can be computed in polynomial time over deterministic and decomposable Boolean circuits. Such circuits, also known as tractable Boolean circuits, generalize a wide range of Boolean circuits and binary decision diagrams classes, including binary decision trees, Ordered Binary Decision Diagrams (OBDDs) and Free Binary Decision Diagrams (FBDDs). We also establish the computational limits of the notion of SHAP-score by observing that, under a mild condition, computing it over a class of Boolean models is always polynomially as hard as the model counting problem for that class. This implies that both determinism and decomposability are essential properties for the circuits that we consider, as removing one or the other renders the problem of computing the SHAP-score intractable (namely, #P-hard)., Comment: 17 pages, including 8 pages of main text. arXiv version of the AAAI'21 conference paper. Except from the addition of the technical appendix, the content is the same as the AAAI one

We consider the problem of exact probabilistic inference for Union of Conjunctive Queries (UCQs) on tuple-independent databases. For this problem, two approaches currently coexist. In the extensional method, query evaluation is performed by exploiting the structure of the query, and relies heavily on the use of the inclusion-exclusion principle. In the intensional method, one first builds a representation of the lineage of the query in a tractable formalism of knowledge compilation. The chosen formalism should then ensure that the probability can be efficiently computed using simple disjointness and independence assumptions, without the need of performing inclusion-exclusion. The extensional approach has long been thought to be strictly more powerful than the intensional approach, the reason being that for some queries, the use of inclusion-exclusion seemed unavoidable. In this paper we introduce a new technique to construct lineage representations as deterministic decomposable circuits in polynomial time. We prove that this technique applies to a class of UCQs that had been conjectured to separate the complexity of the two approaches. In essence, we show that relying on the inclusion-exclusion formula can be avoided by using negation. This result brings back hope to prove that the intensional approach can handle all tractable UCQs., Comment: 16 pages, including 12 pages of main text. This is the ArXiv version of the PODS'20 paper. Except from minor differences that could be introduced by the publisher, the only difference should be the addition of the figure in Appendix D

There exist two approaches for exact probabilistic inference of UCQs on tuple-independent databases. In the extensional approach, query evaluation is performed within a DBMS by exploiting the structure of the query. In the intensional approach, one first builds a representation of the lineage of the query on the database, then computes the probability of the lineage. In this paper we propose a new technique to construct lineage representations as deterministic decomposable circuits in PTIME. The technique can apply to a class of UCQs that has been conjectured to separate the complexity of the two approaches. We test our technique experimentally, and show that it succeeds on all the queries of this class up to a certain size parameter, i.e., over $20$ million queries., Comment: 10 pages. Appeared in the workshop AMW'18

We study the complexity of various fundamental counting problems that arise in the context of incomplete databases, i.e., relational databases that can contain unknown values in the form of labeled nulls. Specifically, we assume that the domains of these unknown values are finite and, for a Boolean query $q$, we consider the following two problems: given as input an incomplete database $D$, (a) return the number of completions of $D$ that satisfy $q$; or (b) return or the number of valuations of the nulls of $D$ yielding a completion that satisfies $q$. We obtain dichotomies between #P-hardness and polynomial-time computability for these problems when $q$ is a self-join--free conjunctive query, and study the impact on the complexity of the following two restrictions: (1) every null occurs at most once in $D$ (what is called Codd tables); and (2) the domain of each null is the same. Roughly speaking, we show that counting completions is much harder than counting valuations (for instance, while the latter is always in #P, we prove that the former is not in #P under some widely believed theoretical complexity assumption). Moreover, we find that both (1) and (2) reduce the complexity of our problems. We also study the approximability of these problems and show that, while counting valuations always has a fully polynomial randomized approximation scheme, in most cases counting completions does not. Finally, we consider more expressive query languages and situate our problems with respect to known complexity classes., Comment: 29 pages, including 12 pages of main text. This is the arXiv version of the PODS'20 paper. Except from minor differences that could be introduced by the publisher, the only difference should be the addition of the appendix, which contains all the proofs that do not appear in the main text

The field of knowledge compilation establishes the tractability of many tasks by studying how to compile them to Boolean circuit classes obeying some requirements such as structuredness, decomposability, and determinism. However, in other settings such as intensional query evaluation on databases, we obtain Boolean circuits that satisfy some width bounds, e.g., they have bounded treewidth or pathwidth. In this work, we give a systematic picture of many circuit classes considered in knowledge compilation and show how they can be systematically connected to width measures, through upper and lower bounds. Our upper bounds show that bounded-treewidth circuits can be constructively converted to d-SDNNFs, in time linear in the circuit size and singly exponential in the treewidth; and that bounded-pathwidth circuits can similarly be converted to uOBDDs. We show matching lower bounds on the compilation of monotone DNF or CNF formulas to structured targets, assuming a constant bound on the arity (size of clauses) and degree (number of occurrences of each variable): any d-SDNNF (resp., SDNNF) for such a DNF (resp., CNF) must be of exponential size in its treewidth, and the same holds for uOBDDs (resp., n-OBDDs) when considering pathwidth. Unlike most previous work, our bounds apply to any formula of this class, not just a well-chosen family. Hence, we show that pathwidth and treewidth respectively characterize the efficiency of compiling monotone DNFs to uOBDDs and d-SDNNFs with compilation being singly exponential in the corresponding width parameter. We also show that our lower bounds on CNFs extend to unstructured compilation targets, with an exponential lower bound in the treewidth (resp., pathwidth) when compiling monotone CNFs of constant arity and degree to DNNFs (resp., nFBDDs)., Comment: 46 pages. Extended version of arXiv:1709.06188. Up to the stylesheet, page/environment numbering, minor formatting, and publisher-induced changes, this is the exact content of the paper in Theory of Computing Systems . The difference in the titles (missing "and") is an error introduced by the publisher

We investigate parameterizations of both database instances and queries that make query evaluation fixed-parameter tractable in combined complexity. We show that clique-frontier-guarded Datalog with stratified negation (CFG-Datalog) enjoys bilinear-time evaluation on structures of bounded treewidth for programs of bounded rule size. Such programs capture in particular conjunctive queries with simplicial decompositions of bounded width, guarded negation fragment queries of bounded CQ-rank, or two-way regular path queries. Our result is shown by translating to alternating two-way automata, whose semantics is defined via cyclic provenance circuits (cycluits) that can be tractably evaluated., Comment: 56 pages, 63 references. Journal version of "Combined Tractability of Query Evaluation via Tree Automata and Cycluits (Extended Version)" at arXiv:1612.04203. Up to the stylesheet, page/environment numbering, and possible minor publisher-induced changes, this is the exact content of the journal paper that will appear in Theory of Computing Systems. Update wrt version 1: latest reviewer feedback

Several query evaluation tasks can be done via knowledge compilation: the query result is compiled as a lineage circuit from which the answer can be determined. For such tasks, it is important to leverage some width parameters of the circuit, such as bounded treewidth or pathwidth, to convert the circuit to structured classes, e.g., deterministic structured NNFs (d-SDNNFs) or OBDDs. In this work, we show how to connect the width of circuits to the size of their structured representation, through upper and lower bounds. For the upper bound, we show how bounded-treewidth circuits can be converted to a d-SDNNF, in time linear in the circuit size. Our bound, unlike existing results, is constructive and only singly exponential in the treewidth. We show a related lower bound on monotone DNF or CNF formulas, assuming a constant bound on the arity (size of clauses) and degree (number of occurrences of each variable). Specifically, any d-SDNNF (resp., SDNNF) for such a DNF (resp., CNF) must be of exponential size in its treewidth; and the same holds for pathwidth when compiling to OBDDs. Our lower bounds, in contrast with most previous work, apply to any formula of this class, not just a well-chosen family. Hence, for our language of DNF and CNF, pathwidth and treewidth respectively characterize the efficiency of compiling to OBDDs and (d-)SDNNFs, that is, compilation is singly exponential in the width parameter. We conclude by applying our lower bound results to the task of query evaluation., Comment: 33 pages, no figures, 40 references. This is the full version with proofs of the corresponding ICDT'18 publication, and it integrates all reviewer feedback. Except for the additional appendices, and except for formatting differences and inessential changes, the contents are the same as in the conference version. Fixed in version 4 a minor omission in the proof of Theorem 33 and small typos

Query evaluation over probabilistic databases is known to be intractable in many cases, even in data complexity, i.e., when the query is fixed. Although some restrictions of the queries [19] and instances [4] have been proposed to lower the complexity, these known tractable cases usually do not apply to combined complexity, i.e., when the query is not fixed. This leaves open the question of which query and instance languages ensure the tractability of probabilistic query evaluation in combined complexity. This paper proposes the first general study of the combined complexity of conjunctive query evaluation on probabilistic instances over binary signatures, which we can alternatively phrase as a probabilistic version of the graph homomorphism problem, or of a constraint satisfaction problem (CSP) variant. We study the complexity of this problem depending on whether instances and queries can use features such as edge labels, disconnectedness, branching, and edges in both directions. We show that the complexity landscape is surprisingly rich, using a variety of technical tools: automata-based compilation to d-DNNF lineages as in [4], \b{eta}-acyclic lineages using [10], the X-property for tractable CSP from [24], graded DAGs [27] and various coding techniques for hardness proofs., Comment: 36 pages including 4 appendix sections. This is the PODS'17 article with all proofs and all reviewer feedback. Relative to the previous version and to the PODS version, this version adds details about a subtle point in Appendix D, and fixes some minor formatting issues

We investigate parameterizations of both database instances and queries that make query evaluation fixed-parameter tractable in combined complexity. We introduce a new Datalog fragment with stratified negation, intensional-clique-guarded Datalog (ICG-Datalog), with linear-time evaluation on structures of bounded treewidth for programs of bounded rule size. Such programs capture in particular conjunctive queries with simplicial decompositions of bounded width, guarded negation fragment queries of bounded CQ-rank, or two-way regular path queries. Our result proceeds via compilation to alternating two-way automata, whose semantics is defined via cyclic provenance circuits (cycluits) that can be tractably evaluated. Last, we prove that probabilistic query evaluation remains intractable in combined complexity under this parameterization., Comment: 69 pages, accepted at ICDT'17. Appendix F contains results from an independent upcoming journal paper by Michael Benedikt, Pierre Bourhis, Georg Gottlob, and Pierre Senellart

Query answering over probabilistic data is an important task but is generally intractable. However, a new approach for this problem has recently been proposed, based on structural decompositions of input databases, following, e.g., tree decompositions. This paper presents a vision for a database management system for probabilistic data built following this structural approach. We review our existing and ongoing work on this topic and highlight many theoretical and practical challenges that remain to be addressed., Comment: 9 pages, 1 figure, 23 references. Accepted for publication at SUM 2016