Publications (Hide summaries) (Show abstracts) (Hide highlights)

Relational query languages have been studied and used for more than 50 years, with SQL overwhelmingly dominant in practice. Yet that dominance is now being questioned from several directions at once: higher-level abstractions such as entity-relationship and functional data models, application languages that integrate query- ing with application logic, algebraic intermediate representations that blur the boundary between logical and physical specification, and large language models (LLMs) that both generate and explain queries. These developments highlight that relational languages dif- fer not only in expressive power, but also in what relational structure they make explicit and in how effectively they support humans and machines in writing, understanding, revising, and reasoning about queries.
This tutorial takes this moment as an opportunity to rethink relational language design. Rather than beginning from formal def- initions, we start from a fixed set of representative queries and compare how SQL, classical alternatives, and several recent and earlier languages express the same intent. We organize this com- parison around four primary aspects that are often conflated: query intent, relational pattern, encoding style, and semantic conventions. From these examples, we derive a common vocabulary of recur- ring design dimensions and trade-offs in relational language design. Participants will leave with clearer mental models for comparing existing and proposed languages, evaluating their usability for hu- mans and AI systems, and articulating open problems in the design and evaluation of relational query languages.
The tutorial webpage is at: https://northeastern-datalab.github.i o/relational-language-tutorial.

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).

In recent years, there has been significant progress in the development of deep learning models over relational databases, including architectures based on heterogeneous graph neural networks (hetero-GNNs) and heterogeneous graph transformers. In effect, such architectures state how the database records and links (e.g., foreign-key references) translate into a large, complex numerical expression, involving numerous learnable parameters. This complexity makes it hard to explain, in human-understandable terms, how a model uses the available data to arrive at a given prediction.
We present a novel framework for explaining machine-learning models over relational databases, where explanations are view definitions that highlight focused parts of the database that mostly contribute to the model's prediction. We establish such global abductive explanations by adapting the classic notion of determinacy by Nash, Segoufin, and Vianu (2010). In addition to tuning the tradeoff between determinacy and conciseness, the framework allows controlling the level of granularity by adopting different fragments of view definitions, such as ones highlighting whole columns, foreign keys between tables, relevant groups of tuples, and so on.
We investigate the realization of the framework in the case of hetero-GNNs. We develop heuristic algorithms that avoid the exhaustive search over the space of all databases. We propose techniques that are model-agnostic, and others that are tailored to hetero-GNNs via the notion of learnable masking. Our approach is evaluated through an extensive empirical study on the RelBench collection, covering a variety of domains and different record-level tasks. The results demonstrate the usefulness of the proposed explanations, as well as the efficiency of their generation.

Filters are ubiquitous in computer science, enabling space-efficient approximate membership testing. Since Bloom filters were introduced in 1970, decades of work improved their space efficiency and performance. Recently, three new paradigms have emerged offering orders-of-magnitude improvements in false positive rates (FPRs) by using information beyond the input set: (1) learned filters train a model to distinguish (non)members, (2) stacked filters use negative workload samples to build cascading layers, and (3) adaptive filters update internal representation in response to false positive feedback. Yet each paradigm targets specific use cases, introduces complex configuration tuning, and has been evaluated in isolation. This results in unclear trade-offs and a gap in understanding how these approaches compare and when each is most appropriate.
This paper presents the first comprehensive evaluation of learned, stacked, and adaptive filters across real-world datasets and query workloads. Our results reveal critical trade-offs: (1) Learned filters achieve up to 10^2 times lower FPRs but exhibit high variance and lack robustness under skewed or dynamic workloads. Critically, model inference overhead leads to up to 10^4 times slower query latencies than stacked or adaptive filters. (2) Stacked filters reliably achieve up to 10^3 times lower FPRs on skewed workloads but require workload knowledge. (3) Adaptive filters are robust across settings, achieving up to 10^3 times lower FPRs under adversarial queries without workload assumptions. Based on our analysis, learned filters suit stable workloads where input features enable effective model training and space constraints are paramount, stacked filters excel when reliable query distributions are known, and adaptive filters are most generalizable, providing robust theoretically bound guarantees even in dynamic or adversarial environments.

We demonstrate PatternVis, an interactive tool that—given a SQL query and its schema—can represent the relational pattern of the query in an extended representation of Relational Diagrams. Notably, PatternVis can represent groupings and aggregation in addition to the non-disjunctive fragment of Relational Diagrams.
Our interactive demo has two main contributions: (1) To the best of our knowledge, this is the first interactive SQL visualization tool that can give an unambiguous representation of nested SQL queries with GROUP BY clauses and aggregation. (2) We present a set of tasks involving multiple-choice questions in which the goal is to identify the correct output of a given aggregation query and database instance. Users can choose to complete each task in one of three modalities (SQL only, PatternVis only, or both) and experience the varying difficulties in each modality.

The resilience problem for a query and an input set or bag database is to compute the minimum number of facts to remove from the database to make the query false. In this paper, we study how to compute the resilience of Regular Path Queries (RPQs) over graph databases. Our goal is to characterize the regular languages 𝐿 for which it is tractable to compute the resilience of the existentially-quantified RPQ built from 𝐿.
We show that computing the resilience in this sense is tractable (even in combined complexity) for all RPQs defined from so-called local languages. By contrast, we show hardness in data complexity for RPQs defined from the following language classes (after reducing the languages to eliminate redundant words): all finite languages featuring a word containing a repeated letter, and all languages featuring a specific kind of counterexample to being local (which we call four-legged languages). The latter include in particular all languages that are not star-free. Our results also imply hardness for all non-local languages with a so-called neutral letter.
We also highlight some remaining obstacles towards a full dichotomy. In particular, for the RPQ 𝑎𝑏𝑐|𝑏𝑒, resilience is tractable but the only PTIME algorithm that we know uses submodular function optimization.

We consider learning a probabilistic classifier from partially-labelled supervision (inputs denoted with multiple possibilities) using standard neural architectures with a softmax as the final layer. We identify a bias phenomenon that can arise from the softmax layer in even simple architectures that prevents proper exploration of alternative options, making the dynamics of gradient descent overly sensitive to initialisation. We introduce a novel loss function that allows for unbiased exploration within the space of alternative outputs. We give a theoretical justification for our loss function, and provide an extensive evaluation of its impact on synthetic data, on standard partially labelled benchmarks and on a contributed novel benchmark related to an existing rule learning challenge.

Large enterprise databases can be complex and messy, obscuring the data semantics needed for analytical tasks. We propose a semantic layer in-between the database and the user as a set of small and easy-to-interpret database views, effectively acting as a refined version of the schema. To discover these views, we introduce a multi-agent Large Language Model (LLM) simulation where LLM agents collaborate to iteratively define and refine views with minimal input. Our approach paves the way for LLM-powered exploration of unwieldy databases.

Query formulation is increasingly performed by systems that need to guess a user’s intent (e.g. via spoken word interfaces). But how can a user know that the computational agent is returning answers to the “right” query? More generally, given that relational queries can become pretty complicated, how can we help users understand relational queries, whether human-generated or automatically generated? Now seems the right moment to revisit a topic that predates the birth of the relational model: developing visual metaphors that help users understand relational queries.
This lecture-style tutorial surveys the key visual metaphors developed for diagrammatic representations of logical statements and relational expressions, across both the relational database and the much older diagrammatic reasoning communities. We survey the history and state-of-the-art of relationally-complete diagrammatic representations of relational queries, discuss the key visual metaphors developed in over a century of investigations into diagrammatic languages, and organize the landscape by mapping the visual alphabets of diagrammatic representation systems to the syntax and semantics of Relational Algebra (RA) and Relational Calculus (RC). Tutorial website: https://northeastern-datalab.github.io/diagrammatic-representation-tutorial/

We analyze a general problem in a crowd-sourced setting where one user asks a question (also called item) and other users return answers (also called labels) for this question. Different from existing crowd sourcing work which focuses on finding the most appropriate label for the question (the “truth”), our problem is to determine a ranking of the users based on their ability to answer questions. We call this problem “ability discovery” to emphasize the connection to and duality with the more well-studied problem of “truth discovery”.
To model items and their labels in a principled way, we draw upon Item Response Theory (IRT) which is the widely accepted theory behind standardized tests such as SAT and GRE. We start from an idealized setting where the relative performance of users is consistent across items and better users choose better fitting labels for each item. We posit that a principled algorithmic solution to our more general problem should solve this ideal setting correctly and observe that the response matrices in this setting obey the Consecutive Ones Property (C1P). While C1P is well understood algorithmically with various discrete algorithms, we devise a novel variant of the HITS algorithm which we call “HITSnDIFFs” (or HnD), and prove that it can recover the ideal C1P-permutation in case it exists. Unlike fast combinatorial algorithms for finding the consecutive ones permutation (if it exists), H N D also returns an ordering when such a permutation does not exist. Thus it provides a principled heuristic for our problem that is guaranteed to return the correct answer in the ideal setting. Our experiments show that H N D produces user rankings with robustly high accuracy compared to state-of-the-art truth discovery methods. We also show that our novel variant of HITS scales better in the number of users than ABH, the only prior spectral C1P reconstruction algorithm.

Query formulation is increasingly performed by systems that need to guess a user’s intent (e.g. via spoken word interfaces). But how can a user know that the computational agent is returning answers to the “right” query? More generally, given that relational queries can become pretty complicated, how can we help users understand existing relational queries, whether human-generated or automatically generated? Now seems the right moment to revisit a topic that predates the birth of the relational model: developing visual metaphors that help users understand relational queries.
This lecture-style tutorial surveys the key visual metaphors developed for visual representations of relational expressions. We will survey the history and state-of-the art of relationally-complete diagrammatic representations of relational queries, discuss the key visual metaphors developed in over a century of investigating diagrammatic languages, and organize the landscape by mapping their used visual alphabets to the syntax and semantics of Relational Algebra (RA) and Relational Calculus (RC).

Modern data lakes are deeply heterogeneous in the vocabulary that is used to describe data. We study a problem of disambiguation in data lakes: how can we determine if a data value occurring more than once in the lake has different meanings and is therefore a homograph? While word and entity disambiguation have been well studied in computational linguistics, data management and data science, we show that data lakes provide a new opportunity for disambiguation of data values since they represent a massive network of interconnected values. We investigate to what extent this network can be used to disambiguate values.
DomainNet uses network-centrality measures on a bipartite graph whose nodes represent values and attributes to determine, without supervision, if a value is a homograph. A thorough experimental evaluation demonstrates that state-of-the-art techniques in domain discovery cannot be re-purposed to compete with our method. Specifically, using a domain discovery method to identify homographs has a precision and a recall of 38% versus 69% with our method on a synthetic benchmark. By applying a network-centrality measure to our graph representation, DomainNet achieves a good separation between homographs and data values with a unique meaning. On a real data lake our top200 precision is 89%.

Existing techniques for unionable table search define unionability using metadata (tables must have the same or similar schemas) or column-based metrics (for example, the values in a table should be drawn from the same domain). In this work, we introduce the use of semantic relationships between pairs of columns in a table to improve the accuracy of union search. Consequently, we introduce a new notion of unionability that considers relationships between columns, together with the semantics of columns, in a principled way. To do so, we present two new methods to discover semantic relationship between pairs of columns: The first uses an existing knowledge base (KB), the second (which we call a “synthesized KB”) uses knowledge from the data lake itself. We adopt an existing Table Union Search benchmark and present new (open) benchmarks that represent small and large real data lakes. We show that our new unionability search algorithm called SANTOS outperforms a state- of-the-art union search that uses a wide variety of column-based semantics, including word embeddings and regular expressions. We show empirically in all benchmarks that our synthesized KB improves the accuracy of union search by representing relationship semantics that may not be contained in an available KB. This result hints at a promising future of creating a synthesized KB from the data lakes having limited KB coverage and perform union search over them.

We present efficient algorithms for Quantile Join Queries, abbreviated as %JQ. A %JQ asks for the answer at some specified relative position (e.g., 50% for the median) under some ordering over the answers to a Join Query (JQ). Our goal is to avoid materializing the set of all query answers, and to achieve quasilinear time in the size of the database, regardless of the total number of answers. A recent dichotomy result rules out the existence of such an algorithm for a general family of queries and orders. Specifically, for acyclic JQs without self-joins, the problem becomes intractable for ordering by sum whenever we are essentially joining more than two relations. Moreover, even for basic ranking functions beyond sum, such as min or max over different attributes, it has not been known so far whether there is any nontrivial tractable %JQ.
In this work, we develop a new approach to solving %JQ and show how this approach allows not just to recover known results, but also generalize them and resolve open cases. Our solution uses two subroutines: The first one needs to select what we call a ``pivot answer.'' The second subroutine partitions the space of query answers according to this pivot, and continues searching in one partition that is represented as new %JQ over a new database. For pivot selection, we develop an algorithm that, under mild assumptions of monotonicity, works for every ranking function. The second subroutine requires a customized construction for the specific ranking function at hand.
We show the benefit and generality of our approach by using it to establish several new complexity results. First, we prove the tractability of min and max for all acyclic JQs, thereby resolving the above question. Second, we extend the previous dichotomy for sum to all partial sums (over all subsets of the attributes). Third, we handle the intractable cases of sum by devising a deterministic approximation scheme that applies to every acyclic JQ.

We study the question of when we can answer a Conjunctive Query (CQ) with an ordering over the answers by constructing a structure for direct (random) access to the sorted list of answers, without actually materializing this list, so that the construction time is linear (or quasilinear) in the size of the database. In the absence of answer ordering, such a construction has been devised for the task of enumerating query answers of free-connex acyclic CQs, so that the access time is logarithmic. Moreover, it follows from past results that within the class of CQs without self-joins, being free-connex acyclic is necessary for the existence of such a construction (under conventional assumptions in fine-grained complexity).
In this work, we embark on the challenge of identifying the answer orderings that allow for ranked direct access with the above complexity guarantees. We begin with the class of lexicographic orderings and give a decidable characterization of the class of feasible such orderings for every CQ without self-joins. We then continue to the more general case of orderings by the sum of attribute scores. As it turns out, in this case ranked direct access is feasible only in trivial cases. Hence, to better understand the computational challenge at hand, we consider the more modest task of providing access to only one single answer (i.e., finding the answer at a given position). We indeed achieve a quasilinear-time algorithm for a subset of the class of full CQs without self-joins, by adopting a solution of Frederickson and Johnson to the classic problem of selection over sorted matrices. We further prove that none of the other queries in this class admit such an algorithm.

We propose Higher-Order Networks (HONs) for historically challenging problems for Graph Neural Networks (GNNs), such as Constraint Satisfaction Problems (CSPs). We apply a simple extension of GNNs to HONs and show its advantages for solving 3-coloring.

We have made tremendous strides in providing tools for data scientists to discover new tables useful for their analyses. But despite these advances, the proper integration of discovered tables has been under-explored. An interesting semantics for integration, called Full Disjunction, was proposed in the 1980’s, but there has been little progress in using it for data science to integrate tables culled from data lakes. We provide ALITE, a new proposal for integration of tables that may have been discovered using join, union or related table search. We empirically show that ALITE can outperform previous algorithms for computing the Full Disjunction. ALITE relaxes previous assumptions that tables share common attribute names (which completely determine the join columns), are complete (without null values), and have acyclic join patterns. To evaluate ALITE, we develop and share three new benchmarks for integration that use real data lake tables.

Query Visualization (QV) is the problem of transforming a given query into a graphical representation that helps humans understand its meaning. This task is notably different from designing a Visual Query Language (VQL) that helps a user compose a query. This article discusses the principles of relational query visualization and its potential for simplifying user interactions with relational data.

When processing join queries over big data, a DBMS can become unresponsive, i.e., it takes very long until any output tuples appear. Ranked enumeration addresses this problem by attempting to return the most important answers as quickly as possible, ideally in time that is linear (or quasilinear) in input size, even if the complete output is much larger. Aside from its practical usefulness, ranked enumeration is closely related to, and in a way unifies, several other problems involving joins. The common goal is the design of optimal algorithms that are guaranteed to avoid large intermediate results and thus achieve time or space complexity close to a lower bound. Arguably, avoiding query plans that produce huge intermediate results has been an overarching goal of database optimizers, which is part of the reason why optimal join algorithms, enumeration, and factorized representations have generated a lot of excitement. In this tutorial, we embark on an exploration of these topics, showing how they are intimately connected with a wide range of fundamental problems in computer science.

Modern data lakes are deeply heterogeneous in the vocabulary that is used to describe data. We study a problem of disambiguation in data lakes: how can we determine if a data value occurring more than once in the lake has different meanings and is therefore a homograph? While word and entity disambiguation have been well studied in computational linguistics, data management and data science, we show that data lakes provide a new opportunity for disambiguation of data values since they represent a massive network of interconnected values. We investigate to what extent this network can be used to disambiguate values.
DomainNet uses network-centrality measures on a bipartite graph whose nodes represent values and attributes to determine, without supervision, if a value is a homograph. A thorough experimental evaluation demonstrates that state-of-the-art techniques in domain discovery cannot be re-purposed to compete with our method. Specifically, using a domain discovery method to identify homographs has a precision and a recall of 38% versus 69% with our method on a synthetic benchmark. By applying a network-centrality measure to our graph representation, DomainNet achieves a good separation between homographs and data values with a unique meaning. On a real data lake our top200 precision is 89%.

Node-link visualizations are a familiar and powerful tool for displaying the relationships in a network. The readability of these visualizations highly depends on the spatial layout used for the nodes. In this paper, we focus on computing layered layouts, in which nodes are aligned on a set of parallel axes to better expose hierarchical or sequential relationships. Heuristic-based layouts are widely used as they scale well to larger networks and usually create readable, albeit sub-optimal, visualizations. We instead use a layout optimization model that prioritizes optimality --- as compared to scalability --- because an optimal solution not only represents the best attainable result, but can also serve as a baseline to evaluate the effectiveness of layout heuristics. We take an important step towards powerful and flexible network visualization by proposing STRATIFIMAL LAYOUT, a modular integer-linear-programming formulation that can consider several important readability criteria simultaneously --- crossing reduction, edge bendiness, and nested and multi-layer groups. The layout can be adapted to diverse use cases through its modularity. Individual features can be enabled and customized depending on the application. We provide open-source and documented implementations of the layout, both for web-based and desktop visualizations. As a proof-of-concept, we apply it to the problem of visualizing complicated SQL queries, which have features that we believe cannot be addressed by existing layout optimization models. We also include a benchmark network generator and the results of an empirical evaluation to assess the performance trade-offs of our design choices.

Top-k queries have been studied intensively in the database community and they are an important means to reduce query cost when only the “best” or “most interesting” results are needed instead of the full output. While some optimality results exist, e.g., the famous Threshold Algorithm, they hold only in a fairly limited model of computation that does not account for the cost incurred by large intermediate results and hence is not aligned with typical database-optimizer cost models. On the other hand, the idea of avoiding large intermediate results is arguably the main goal of recent work on optimal join algorithms, which uses the standard RAM model of computation to determine algorithm complexity. This research has created a lot of excitement due to its promise of reducing the time complexity of join queries with cycles, but it has mostly focused on full-output computation. We argue that the two areas can and should be studied from a unified point of view in order to achieve optimality in the common model of computation for a very general class of top-k-style join queries. This tutorial has two main objectives. First, we will explore and contrast the main assumptions, concepts, and algorithmic achievements of the two research areas. Second, we will cover recent, as well as some older, approaches that emerged at the intersection to support efficient ranked enumeration of join-query results. These are related to classic work on k-shortest path algorithms and more general optimization problems, some of which dates back to the 1950s. We demonstrate that this line of research warrants renewed attention in the challenging context of ranked enumeration for general join queries.

We consider running-time optimization for band-joins in a distributed system, e.g., the cloud. To balance load across worker machines, input has to be partitioned, which causes duplication. We explore how to resolve this tension between maximum load per worker and input duplication for band-joins between two relations. Previous work suffered from high optimization cost or considered partitionings that were too restricted (resulting in suboptimal join performance).
Our main insight is that recursive partitioning of the join-attribute space with the appropriate split scoring measure can achieve both low optimization cost and low join cost. It is the first approach that is not only effective for one-dimensional band-joins but also for joins on multiple attributes. Experiments indicate that our method is able to find partitionings that are within 10% of the lower bound for both maximum load per worker and input duplication for a broad range of settings, significantly improving over previous work.

We introduce a new set of problems based on the Chain Editing problem. In our version of Chain Editing, we are given a set of participants and a set of tasks that every participant attempts. For each participant-task pair, we know whether the participant has succeeded at the task or not. We assume that participants vary in their ability to solve tasks, and that tasks vary in their difficulty to be solved. In an ideal world, stronger participants should succeed at a superset of tasks that weaker participants succeed at. Similarly, easier tasks should be completed successfully by a superset of participants who succeed at harder tasks. In reality, it can happen that a stronger participant fails at a task that a weaker participants succeeds at. Our goal is to find a perfect nesting of the participant-task relations by flipping a minimum number of participant-task relations, implying such a "nearest perfect ordering" to be the one that is closest to the truth of participant strengths and task difficulties. Many variants of the problem are known to be NP-hard.

Many problems in areas as diverse as recommendation systems, social network analysis, semantic search, and distributed root cause analysis can be modeled as pattern search on labeled graphs (also called "heterogeneous information networks" or HINs). Given a large graph and a query pattern with node and edge label constraints, a fundamental challenge is to find the top-k matches according to a ranking function over edge and node weights. For users, it is difficult to select value k. We therefore propose the novel notion of an any-k ranking algorithm: for a given time budget, return as many of the top-ranked results as possible. Then, given additional time, produce the next lower-ranked results quickly as well. It can be stopped anytime, but may have to continue until all results are returned. This paper focuses on acyclic patterns over arbitrary labeled graphs. We are interested in practical algorithms that effectively exploit (1) properties of heterogeneous networks, in particular selective constraints on labels, and (2) that the users often explore only a fraction of the top-ranked results. Our solution, KARPET, carefully integrates aggressive pruning that leverages the acyclic nature of the query, and incremental guided search. It enables us to prove strong non-trivial time and space guarantees, which is generally considered very hard for this type of graph search problem. Through experimental studies we show that KARPET achieves running times in the order of milliseconds for tree patterns on large networks with millions of nodes and edges.

We recently proposed the notion of any-k queries, together with the KARPET algorithm, for tree-pattern search in labeled graphs. Anyk extends top-k by not requiring a pre-specified value of k. Instead, an any-k algorithm returns as many of the top-ranked results as possible, for a given time budget. Given additional time, it produces the next-highest ranked results quickly as well. It can be stopped anytime, but may have to continue until all results are returned. In the latter case, any-k takes times similar to an algorithm that first produces all results and then sorts them. We summarize KARPET and argue that it can be extended to support any-k exploratory search for arbitrary conjunctive queries.

Tuple-independent and disjoint-independent probabilistic databases (TI- and DI-PDBs) represent uncertain data in a factorized form as a product of independent random variables that represent either tuples (TI-PDBs) or sets of tuples (DI-PDBs). When the user submits a query, the database derives the marginal probabilities of each output-tuple, exploiting the underlying assumptions of statistical independence. While query processing in TI- and DI-PDBs has been studied extensively, limited research has been dedicated to the problems of updating or deriving the parameters from observations of query results. Addressing this problem is the main focus of this article. We first introduce Beta Probabilistic Databases (B-PDBs), a generalization of TI-PDBs designed to support both (i) belief updating and (ii) parameter learning in a principled and scalable way. The key idea of B-PDBs is to treat each parameter as a latent, Beta-distributed random variable. We show how this simple expedient enables both belief updating and parameter learning in a principled way, without imposing any burden on regular query processing. Building on B-PDBs, we then introduce Dirichlet Probabilistic Databases (D-PDBs), a generalization of DI-PDBs with similar properties. We provide the following key contributions for both B- and D-PDBs: (i) We study the complexity of performing Bayesian belief updates and devise efficient algorithms for certain tractable classes of queries; (ii) we propose a soft-EM algorithm for computing maximum-likelihood estimates of the parameters; (iii) we present an algorithm for efficiently computing conditional probabilities, allowing us to efficiently implement B- and D-PDBs via a standard relational engine; and (iv) we support our conclusions with extensive experimental results.

Anytime approximation algorithms that compute the probabilities of queries over probabilistic databases can be of great use to statistical learning tasks. Those approaches have been based so far on either (i) sampling or (ii) branch-and-bound with model-based bounds. We present here a more general branch-and-bound framework that extends the possible bounds by using "dissociation," which yields tighter bounds.

Detecting conflicts of interest (COIs) is key for guaranteeing the fairness of a peer-review process. In many conference management systems, the COIs of authors and reviewers are self-declared, and the declaration process is time consuming and potentially incomplete. To address this problem, we demonstrate a novel interactive system called PISTIS that assists the declaration process in a semi-automatic manner. Apart from keyword search and simple filtering, our system provides an interactive graphical interface that helps users explore potential COIs based on the heterogenous data sources. To simply the process of declaration, we also recommend latent COIs using a supervised ranking model that can be iteratively refined from the data collected from past declarations. We believe that PISTIS can be useful as an assistant tool in many real world conference management systems

Tuple-independent probabilistic databases (TI-PDBs) handle uncertainty by annotating each tuple with a probability parameter; when the user submits a query, the database derives the marginal probabilities of each output-tuple, assuming input-tuples are statistically independent. While query processing in TI-PDBs has been studied extensively, limited research has been dedicated to the problems of updating or deriving the parameters from observations of query results. Addressing this problem is the main focus of this paper. We introduce Beta Probabilistic Databases (B-PDBs), a generalization of TI-PDBs designed to support both (i) belief updating and (ii) parameter learning in a principled and scalable way. The key idea of B-PDBs is to treat each parameter as a latent, Beta-distributed random variable. We show how this simple expedient enables both belief updating and parameter learning in a principled way, without imposing any burden on regular query processing. We use this model to provide the following key contributions: (i) we show how to scalably compute the posterior densities of the parameters given new evidence; (ii) we study the complexity of performing Bayesian belief updates, devising efficient algorithms for tractable classes of queries; (iii) we propose a soft-EM algorithm for computing maximum-likelihood estimates of the parameters; (iv) we show how to embed the proposed algorithms into a standard relational engine; (v) we support our conclusions with extensive experimental results.

Belief Propagation (BP) is a widely used approximation for exact probabilistic inference in graphical models, such as Markov Random Fields (MRFs). In graphs with cycles, however, no exact convergence guarantees for BP are known, in general. For the case when all edges in the MRF carry the same symmetric, doubly stochastic potential, recent works have proposed to approximate BP by linearizing the update equations around default values, which was shown to work well for the problem of node classification. The present paper generalizes all prior work and derives an approach that approximates loopy BP on any pairwise MRF with the problem of solving a linear equation system. This approach combines exact convergence guarantees and a fast matrix implementation with the ability to model heterogenous networks. Experiments on synthetic graphs with planted edge potentials show that the linearization has comparable labeling accuracy as BP for graphs with weak potentials, while speeding-up inference by orders of magnitude.

Peer review is the most critical process in evaluating an article to be accepted for publication in an academic venue. When assigning a reviewer to evaluate an article, the assignment should be aware of conflicts of interest (COIs) such that the reviews are fair to everyone. However, existing conference management systems simply ask reviewers and authors to declare their explicit COIs through a plain search user interface guided by some simple conflict rules. We argue that such declaration system is not enough to discover all latent COI cases. In this work, we study a graphical declaration system that visualizes the relationships of authors and reviewers based on a heterogeneous co-authorship network. With the help of the declarations, we attempt to detect the latent COIs automatically based on the meta-paths of a heterogeneous network.

We consider the weighted model counting task which includes important tasks in graphical models, such as computing the partition function and probability of evidence as special cases. We propose a novel partition-based bounding algorithm that exploits logical structure and gives rise to a set of inequalities from which upper (or lower) bounds can be derived efficiently. The bounds come with optimality guarantees under certain conditions and are oblivious in that they require only limited observations of the structure and parameters of the problem. We experimentally compare our bounds with the mini-bucket scheme (which is also oblivious) and show that our new bounds are often superior and never worse on a wide variety of benchmark networks.

We propose a novel linear semi-supervised learning formulation that is derived from a solid probabilistic framework: belief propagation. We show that our formulation generalizes a number of label propagation algorithms described in the literature by allowing them to propagate generalized assumptions about influences between classes of neighboring nodes. We call this formulation Semi-Supervised Learning with Heterophily (SSL-H). We also show how the modularization matrix can be learned from observed data with a simple convex optimization framework that is inspired by locally linear embedding. We call this approach Linear Heterophily Estimation (LHE). Experiments on synthetic data show that both approaches combined can learn heterophily of agraph with 1M nodes and 10M edges in under 1min.

Within the context of the ongoing deregulation of the electricity industry, we disprove in the first part the commonly stated assumption that, in theory and under the condition of perfect information, decentralized and centralized unit commitment would lead to the same power quantities traded and, hence, to the same optimal social welfare. We show that, even in the absence of any uncertainties, independent optimization of the individual performance objectives by the decentralized market participants can actually lead to lower efficiency than centralized minimization of total operating cost. This result concerns short-term supply optimization for a given demand, and does not consider long-term investment issues. In the second part, we take the position of an individual market participant in a deregulated electricity market. We investigate the task of optimally self-scheduling generators to maximize profits by using stochastic dynamic programming. With the help of actual price forecast data from the ISO New England electricity wholesale market, we demonstrate the improvements that result from modeling the forecast price errors assuming a Cauchy error distribution instead of the commonly used Normal distribution.

This paper shows how causal reasoning can be used for post-factum data cleaning, where errors are detected after data has been transformed (e.g. by a query) and which need to be corrected in the original input data. We achieve this with a novel way for translating the problem into a SAT and a weighted MAX-SAT problem, and then using very effective existing tools.

Visual Query Languages study ways to help users compose queries with visual metaphors. Information Visualization studies automatic visualization techniques to help users understand and analyze data. Query Management focuses on ways to help users manage and re-use existing queries. We observe that there is a related research question across those three topics which has not received much attention, namely that of Query Visualization: How to visually represent a query to help users quickly understand its intent? Here we argue that the involved challenges are still markedly different from those of the other three, that a solution can considerably improve the usability of DBMSs, and that the topic is thus worthy of attention. We envision, that in a few years, there will be free, modular, and lightweight tools available that allow users to visualize and interpret their queries.

WebPageDump is a Firefox extension which allows you to save local copies of pages from the Web. It sounds simple, but it's not. The standard "Save page as" function of web browsers fails with most dynamic web pages and this shortcoming was a serious problem for our research. Comes the birth of WebPageDump. We hope you find it useful too.

Within the context of the ongoing deregulation of the electricity industry, we disprove in the first part the commonly stated assumption that, in theory and under the condition of perfect information, decentralized and centralized unit commitment would lead to the same power quantities traded and, hence, to the same optimal social welfare. We show that, even in the absence of any uncertainties, independent optimization of the individual performance objectives by the decentralized market participants can actually lead to lower efficiency than centralized minimization of total operating cost. This result concerns short-term supply optimization for a given demand, and does not consider long-term investment issues.
In the second part, we take the position of an individual market participant in a deregulated electricity market. We investigate the task of optimally self-scheduling generators to maximize profits by using stochastic dynamic programming. With the help of actual price forecast data from the ISO New England electricity wholesale market, we demonstrate the improvements that result from modeling the forecast price errors assuming a Cauchy error distribution instead of the commonly used Normal distribution.
Finally, by taking statistic uncertainties and inter-temporal interdependencies into account, we show in the third part that a generator owner's optimum bid sequence for a centralized auction market is generally above marginal cost. In contrast to current literature, this is true even where absolutely no abuse of market power is involved. We conclude that marginal production costs cannot be used by themselves as baseline for the assessment of market power in electricity markets.