Showing papers on "Tuple published in 2017"
Posted Content•
TL;DR: It is shown that there is a meaningful gap between the human and machine performances, which suggests that the proposed dataset could well serve as a benchmark for question-answering.
Abstract: We publicly release a new large-scale dataset, called SearchQA, for machine comprehension, or question-answering. Unlike recently released datasets, such as DeepMind CNN/DailyMail and SQuAD, the proposed SearchQA was constructed to reflect a full pipeline of general question-answering. That is, we start not from an existing article and generate a question-answer pair, but start from an existing question-answer pair, crawled from J! Archive, and augment it with text snippets retrieved by Google. Following this approach, we built SearchQA, which consists of more than 140k question-answer pairs with each pair having 49.6 snippets on average. Each question-answer-context tuple of the SearchQA comes with additional meta-data such as the snippet's URL, which we believe will be valuable resources for future research. We conduct human evaluation as well as test two baseline methods, one simple word selection and the other deep learning based, on the SearchQA. We show that there is a meaningful gap between the human and machine performances. This suggests that the proposed dataset could well serve as a benchmark for question-answering.
450 citations
Posted Content•
TL;DR: This paper proposed a new loss function called generalized end-to-end (GE2E) loss, which makes the training of speaker verification models more efficient than their previous tuple-based end to end (TE2E), which does not require an initial stage of example selection.
Abstract: In this paper, we propose a new loss function called generalized end-to-end (GE2E) loss, which makes the training of speaker verification models more efficient than our previous tuple-based end-to-end (TE2E) loss function. Unlike TE2E, the GE2E loss function updates the network in a way that emphasizes examples that are difficult to verify at each step of the training process. Additionally, the GE2E loss does not require an initial stage of example selection. With these properties, our model with the new loss function decreases speaker verification EER by more than 10%, while reducing the training time by 60% at the same time. We also introduce the MultiReader technique, which allows us to do domain adaptation - training a more accurate model that supports multiple keywords (i.e. "OK Google" and "Hey Google") as well as multiple dialects.
239 citations
TL;DR: This paper presents a meta-analyses of nonlinear Analysis and Applied Mathematics (NAAM) and its applications in telecommunications systems and networks.
Abstract: Guiwu Weia,b, Mao Lua, Fuad E. Alsaadib, Tasawar Hayatc,d and Ahmed Alsaedid aSchool of Business, Sichuan Normal University, Chengdu, P.R. China bCommunications Systems and Networks (CSN) Research Group, Department of Electrical and Computer Engineering, Faculty of Engineering, King Abdulaziz University, Jeddah, Saudi Arabia cDepartment of Mathematics, Quaid-I-Azam University 45320, Islamabad, Pakistan dNonlinear Analysis and Applied Mathematics (NAAM) Research Group, Department of Mathematics, Faculty of Science, King Abdulaziz University, Jeddah, Saudi Arabia
163 citations
TL;DR: The comparative study of the BMBO with four state-of-the-art classical algorithms clearly points toward the superiority of the former in terms of search accuracy, convergent capability and stability in solving the 0–1 KP, especially for the high-dimensional instances.
Abstract: This paper presents a novel binary monarch butterfly optimization (BMBO) method, intended for addressing the 0–1 knapsack problem (0–1 KP). Two tuples, consisting of real-valued vectors and binary vectors, are used to represent the monarch butterfly individuals in BMBO. Real-valued vectors constitute the search space, whereas binary vectors form the solution space. In other words, monarch butterfly optimization works directly on real-valued vectors, while solutions are represented by binary vectors. Three kinds of individual allocation schemes are tested in order to achieve better performance. Toward revising the infeasible solutions and optimizing the feasible ones, a novel repair operator, based on greedy strategy, is employed. Comprehensive numerical experimentations on three types of 0–1 KP instances are carried out. The comparative study of the BMBO with four state-of-the-art classical algorithms clearly points toward the superiority of the former in terms of search accuracy, convergent capability and stability in solving the 0–1 KP, especially for the high-dimensional instances.
123 citations
TL;DR: The notion of QL-operations constructed from tuples, where overlap functions O, grouping functions G and fuzzy negations N are used for the generalization of the implication p → q ≡ ¬ p ∨ ( p ∧ q ) , is introduced.
108 citations
TL;DR: It is shown that any knowledge compilation representations from a class that contain decision-DNNFs can be converted into equivalent Free Binary Decision Diagrams, also known as Read-Once Branching Programs, with only a quasi-polynomial increase in representation size.
Abstract: We prove exponential lower bounds on the running time of the state-of-the-art exact model counting algorithms—algorithms for exactly computing the number of satisfying assignments, or the satisfying probability, of Boolean formulas. These algorithms can be seen, either directly or indirectly, as building Decision-Decomposable Negation Normal Form (decision-DNNF) representations of the input Boolean formulas. Decision-DNNFs are a special case of d-DNNFs where d stands for deterministic. We show that any knowledge compilation representations from a class (called DLDDs in this article) that contain decision-DNNFs can be converted into equivalent Free Binary Decision Diagrams (FBDDs), also known as Read-Once Branching Programs, with only a quasi-polynomial increase in representation size. Leveraging known exponential lower bounds for FBDDs, we then obtain similar exponential lower bounds for decision-DNNFs, which imply exponential lower bounds for model-counting algorithms. We also separate the power of decision-DNNFs from d-DNNFs and a generalization of decision-DNNFs known as AND-FBDDs.We then prove new lower bounds for FBDDs that yield exponential lower bounds on the running time of these exact model counters when applied to the problem of query evaluation in tuple-independent probabilistic databases—computing the probability of an answer to a query given independent probabilities of the individual tuples in a database instance. This approach to the query evaluation problem, in which one first obtains the lineage for the query and database instance as a Boolean formula and then performs weighted model counting on the lineage, is known as grounded inference. A second approach, known as lifted inference or extensional query evaluation, exploits the high-level structure of the query as a first-order formula. Although it has been widely believed that lifted inference is strictly more powerful than grounded inference on the lineage alone, no formal separation has previously been shown for query evaluation. In this article, we show such a formal separation for the first time. In particular, we exhibit a family of database queries for which polynomial-time extensional query evaluation techniques were previously known but for which query evaluation via grounded inference using the state-of-the-art exact model counters requires exponential time.
107 citations
TL;DR: This work presents a novel ER system, called DeepER, that achieves good accuracy, high efficiency, as well as ease-of-use, and requires much less human labeled data and does not need feature engineering, compared with traditional machine learning based approaches.
Abstract: Entity resolution (ER) is a key data integration problem. Despite the efforts in 70+ years in all aspects of ER, there is still a high demand for democratizing ER - humans are heavily involved in labeling data, performing feature engineering, tuning parameters, and defining blocking functions. With the recent advances in deep learning, in particular distributed representation of words (a.k.a. word embeddings), we present a novel ER system, called DeepER, that achieves good accuracy, high efficiency, as well as ease-of-use (i.e., much less human efforts). For accuracy, we use sophisticated composition methods, namely uni- and bi-directional recurrent neural networks (RNNs) with long short term memory (LSTM) hidden units, to convert each tuple to a distributed representation (i.e., a vector), which can in turn be used to effectively capture similarities between tuples. We consider both the case where pre-trained word embeddings are available as well the case where they are not; we present ways to learn and tune the distributed representations. For efficiency, we propose a locality sensitive hashing (LSH) based blocking approach that uses distributed representations of tuples; it takes all attributes of a tuple into consideration and produces much smaller blocks, compared with traditional methods that consider only a few attributes. For ease-of-use, DeepER requires much less human labeled data and does not need feature engineering, compared with traditional machine learning based approaches which require handcrafted features, and similarity functions along with their associated thresholds. We evaluate our algorithms on multiple datasets (including benchmarks, biomedical data, as well as multi-lingual data) and the extensive experimental results show that DeepER outperforms existing solutions.
106 citations
09 May 2017
TL;DR: In this article, the authors consider the problem of enumerating and counting answers to k-ary queries against relational databases that may be updated by inserting or deleting tuples, and show that these can be maintained efficiently in the following sense: during a linear time preprocessing phase, the data structure that enables constant delay enumeration of the query results can be constructed; and when the database is updated, the database can update data structure and restart the enumeration phase within constant time.
Abstract: We consider the task of enumerating and counting answers to k-ary conjunctive queries against relational databases that may be updated by inserting or deleting tuples. We exhibit a new notion of q-hierarchical conjunctive queries and show that these can be maintained efficiently in the following sense. During a linear time pre-processing phase, we can build a data structure that enables constant delay enumeration of the query results; and when the database is updated, we can update the data structure and restart the enumeration phase within constant time. For the special case of self-join free conjunctive queries we obtain a dichotomy: if a query is not q-hierarchical, then query enumeration with sublinear *) delay and sublinear update time (and arbitrary preprocessing time) is impossible.For answering Boolean conjunctive queries and for the more general problem of counting the number of solutions of k-ary queries we obtain complete dichotomies: if the query's homomorphic core is q-hierarchical, then size of the the query result can be computed in linear time and maintained with constant update time. Otherwise, the size of the query result cannot be maintained with sublinear update time.All our lower bounds rely on the OMv-conjecture, a conjecture on the hardness of online matrix-vector multiplication that has recently emerged in the field of fine-grained complexity to characterise the hardness of dynamic problems. The lower bound for the counting problem additionally relies on the orthogonal vectors conjecture, which in turn is implied by the strong exponential time hypothesis.*) By sublinear we mean O(n(1-e) for some e > 0, where n is the size of the active domain of the current database.
98 citations
01 Sep 2017
TL;DR: A query processing model called "relaxed operator fusion" is presented that allows the DBMS to introduce staging points in the query plan where intermediate results are temporarily materialized and reduces the execution time of OLAP queries by up to 2.2× and achieves up to 1.8× better performance compared to other in-memory DBMSs.
Abstract: In-memory database management systems (DBMSs) are a key component of modern on-line analytic processing (OLAP) applications, since they provide low-latency access to large volumes of data. Because disk accesses are no longer the principle bottleneck in such systems, the focus in designing query execution engines has shifted to optimizing CPU performance. Recent systems have revived an older technique of using just-in-time (JIT) compilation to execute queries as native code instead of interpreting a plan. The state-of-the-art in query compilation is to fuse operators together in a query plan to minimize materialization overhead by passing tuples efficiently between operators. Our empirical analysis shows, however, that more tactful materialization yields better performance.We present a query processing model called "relaxed operator fusion" that allows the DBMS to introduce staging points in the query plan where intermediate results are temporarily materialized. This allows the DBMS to take advantage of inter-tuple parallelism inherent in the plan using a combination of prefetching and SIMD vectorization to support faster query execution on data sets that exceed the size of CPU-level caches. Our evaluation shows that our approach reduces the execution time of OLAP queries by up to 2.2× and achieves up to 1.8× better performance compared to other in-memory DBMSs.
95 citations
IBM1
TL;DR: BONIE is designed and released, the first open numerical relation extractor, for extracting Open IE tuples where one of the arguments is a number or a quantity-unit phrase.
Abstract: We design and release BONIE, the first open numerical relation extractor, for extracting Open IE tuples where one of the arguments is a number or a quantity-unit phrase. BONIE uses bootstrapping to learn the specific dependency patterns that express numerical relations in a sentence. BONIE’s novelty lies in task-specific customizations, such as inferring implicit relations, which are clear due to context such as units (for e.g., ‘square kilometers’ suggests area, even if the word ‘area’ is missing in the sentence). BONIE obtains 1.5x yield and 15 point precision gain on numerical facts over a state-of-the-art Open IE system.
91 citations
01 Jul 2017
TL;DR: This work develops a new inference model for Open IE that can work effectively with multiple short facts, noise, and the relational structure of tuples, and significantly outperforms a state-of-the-art structured solver on complex questions of varying difficulty.
Abstract: While there has been substantial progress in factoid question-answering (QA), answering complex questions remains challenging, typically requiring both a large body of knowledge and inference techniques. Open Information Extraction (Open IE) provides a way to generate semi-structured knowledge for QA, but to date such knowledge has only been used to answer simple questions with retrieval-based methods. We overcome this limitation by presenting a method for reasoning with Open IE knowledge, allowing more complex questions to be handled. Using a recently proposed support graph optimization framework for QA, we develop a new inference model for Open IE, in particular one that can work effectively with multiple short facts, noise, and the relational structure of tuples. Our model significantly outperforms a state-of-the-art structured solver on complex questions of varying difficulty, while also removing the reliance on manually curated knowledge.
TL;DR: The problem of computing conjunctive queries over large databases on parallel architectures without shared storage is studied and essentially tight upper and lower bounds for one round algorithms are obtained and how the bounds degrade when there is skew in the data is shown.
Abstract: We study the problem of computing conjunctive queries over large databases on parallel architectures without shared storage. Using the structure of such a query q and the skew in the data, we study tradeoffs between the number of processors, the number of rounds of communication, and the per-processor load—the number of bits each processor can send or can receive in a single round—that are required to compute q. Since each processor must store its received bits, the load is at most the number of bits of storage per processor.When the data are free of skew, we obtain essentially tight upper and lower bounds for one round algorithms, and we show how the bounds degrade when there is skew in the data. In the case of skewed data, we show how to improve the algorithms when approximate degrees of the (necessarily small number of) heavy-hitter elements are available, obtaining essentially optimal algorithms for queries such as skewed simple joins and skewed triangle join queries.For queries that we identify as treelike, we also prove nearly matching upper and lower bounds for multi-round algorithms for a natural class of skew-free databases. One consequence of these latter lower bounds is that for any v > 0, using p processors to compute the connected components of a graph, or to output the path, if any, between a specified pair of vertices of a graph with m edges and per-processor load that is O(m/p1−v) requires Ω(logp) rounds of communication.Our upper bounds are given by simple structured algorithms using MapReduce. Our one-round lower bounds are proved in a very general model, which we call the Massively Parallel Communication (MPC) model, that allows processors to communicate arbitrary bits. Our multi-round lower bounds apply in a restricted version of the MPC model in which processors in subsequent rounds after the first communication round are only allowed to send tuples.
21 Jul 2017
TL;DR: The core of the proposed method is a new co-attention model that learns how to exploit a set of external off-the-shelf algorithms to achieve its goal, an approach that has something in common with the Neural Turing Machine.
Abstract: One of the most intriguing features of the Visual Question Answering (VQA) challenge is the unpredictability of the questions. Extracting the information required to answer them demands a variety of image operations from detection and counting, to segmentation and reconstruction. To train a method to perform even one of these operations accurately from {image, question, answer} tuples would be challenging, but to aim to achieve them all with a limited set of such training data seems ambitious at best. Our method thus learns how to exploit a set of external off-the-shelf algorithms to achieve its goal, an approach that has something in common with the Neural Turing Machine [10]. The core of our proposed method is a new co-attention model. In addition, the proposed approach generates human-readable reasons for its decision, and can still be trained end-to-end without ground truth reasons being given. We demonstrate the effectiveness on two publicly available datasets, Visual Genome and VQA, and show that it produces the state-of-the-art results in both cases.
01 Jan 2017
TL;DR: This paper explains how to use MPI to implement joins, shows the impact and advantages of RDMA, discusses the importance of network scheduling, and study the relative performance of sorting vs. hashing.
Abstract: Traditional database operators such as joins are relevant not only in the context of database engines but also as a building block in many computational and machine learning algorithms. With the advent of big data, there is an increasing demand for efficient join algorithms that can scale with the input data size and the available hardware resources.In this paper, we explore the implementation of distributed join algorithms in systems with several thousand cores connected by a low-latency network as used in high performance computing systems or data centers. We compare radix hash join to sort-merge join algorithms and discuss their implementation at this scale. In the paper, we explain how to use MPI to implement joins, show the impact and advantages of RDMA, discuss the importance of network scheduling, and study the relative performance of sorting vs. hashing. The experimental results show that the algorithms we present scale well with the number of cores, reaching a throughput of 48.7 billion input tuples per second on 4,096 cores.
09 May 2017
TL;DR: A crowd-powered database system CDB is developed that supports crowd-based query optimizations, with focus on join and selection and a unified framework to perform the multi-goal optimization based on the graph model.
Abstract: Crowdsourcing database systems have been proposed to leverage crowd-powered operations to encapsulate the complexities of interacting with the crowd. Existing systems suffer from two major limitations. Firstly, in order to optimize a query, they often adopt the traditional tree model to select an optimized table-level join order. However, the tree model provides a coarse-grained optimization, which generates the same order for different joined tuples and limits the optimization potential that different joined tuples can be optimized by different orders. Secondly, they mainly focus on optimizing the monetary cost. In fact, there are three optimization goals (i.e., smaller monetary cost, lower latency, and higher quality) in crowdsourcing, and it calls for a system to enable multi-goal optimization. To address the limitations, we develop a crowd-powered database system CDB that supports crowd-based query optimizations, with focus on join and selection. CDB has fundamental differences from existing systems. First, CDB employs a graph-based query model that provides more fine-grained query optimization. Second, CDB adopts a unified framework to perform the multi-goal optimization based on the graph model. We have implemented our system and deployed it on AMT, CrowdFlower and ChinaCrowd. We have also created a benchmark for evaluating crowd-powered databases. We have conducted both simulated and real experiments, and the experimental results demonstrate the performance superiority of CDB on cost, latency and quality.
09 May 2017
TL;DR: This paper presents a new approach for evaluating queries under updates that is not only more efficient in terms of memory consumption, but is also consistently faster in processing updates.
Abstract: Modern computing tasks such as real-time analytics require refresh of query results under high update rates. Incremental View Maintenance (IVM) approaches this problem by materializing results in order to avoid recomputation. IVM naturally induces a trade-off between the space needed to maintain the materialized results and the time used to process updates. In this paper, we show that the full materialization of results is a barrier for more general optimization strategies. In particular, we present a new approach for evaluating queries under updates. Instead of the materialization of results, we require a data structure that allows: (1) linear time maintenance under updates, (2) constant-delay enumeration of the output, (3) constant-time lookups in the output, while (4) using only linear space in the size of the database. We call such a structure a Dynamic Constant-delay Linear Representation (DCLR) for the query. We show that DYN, a dynamic version of the Yannakakis algorithm, yields DCLRs for the class of free-connex acyclic CQs. We show that this is optimal in the sense that no DCLR can exist for CQs that are not free-connex acyclic. Moreover, we identify a sub-class of queries for which DYN features constant-time update per tuple and show that this class is maximal. Finally, using the TPC-H and TPC-DS benchmarks, we experimentally compare DYN and a higher-order IVM (HIVM) engine. Our approach is not only more efficient in terms of memory consumption (as expected), but is also consistently faster in processing updates.
09 May 2017
TL;DR: An algebraic framework for describing provenance based on commutative semirings and semimodules over suchSemirings is developed, exploiting usefully the observation that, for database provenance, data use has two flavors: joint and alternative.
Abstract: Imagine a computational process that uses a complex input consisting of multiple "items" (e.g.,files, tables, tuples, parameters, configuration rules) The provenance analysis of such a process allows us to understand how the different input items affect the output of the computation. It can be used, for example, to derive confidence in the output (given confidences in the input items), to derive the minimum access clearance for the output (given input items with different classifications), to minimize the cost of obtaining the output (given a complex input item pricing scheme). It also applies to probabilistic reasoning about an output (given input item distributions), as well as to output maintenance, and to debugging.Provenance analysis for queries, views, database ETL tools, and schema mappings is strongly influenced by their declarative nature, providing mathematically nice descriptions of the output-inputs correlation. In a series of papers starting with PODS 2007 we have developed an algebraic framework for describing such provenance based on commutative semirings and semimodules over such semirings. So far, the framework has exploited usefully the observation that, for database provenance, data use has two flavors: joint and alternative.Here, we have selected several insights that we consider essential for the appreciation of this framework's nature and effectiveness and we also give some idea of its applicability.
Posted Content•
TL;DR: This article proposed a new inference model for Open Information Extraction, in particular one that can work effectively with multiple short facts, noise, and the relational structure of tuples, which significantly outperforms a state-of-the-art structured solver on complex questions.
Abstract: While there has been substantial progress in factoid question-answering (QA), answering complex questions remains challenging, typically requiring both a large body of knowledge and inference techniques. Open Information Extraction (Open IE) provides a way to generate semi-structured knowledge for QA, but to date such knowledge has only been used to answer simple questions with retrieval-based methods. We overcome this limitation by presenting a method for reasoning with Open IE knowledge, allowing more complex questions to be handled. Using a recently proposed support graph optimization framework for QA, we develop a new inference model for Open IE, in particular one that can work effectively with multiple short facts, noise, and the relational structure of tuples. Our model significantly outperforms a state-of-the-art structured solver on complex questions of varying difficulty, while also removing the reliance on manually curated knowledge.
01 Feb 2017
TL;DR: In this paper, the authors propose the Private Data Network (PDN), a federated database for querying over the collective data of mutually distrustful parties, where each member database does not reveal its tuples to its peers nor to the query writer.
Abstract: People and machines are collecting data at an unprecedented rate. Despite this newfound abundance of data, progress has been slow in sharing it for open science, business, and other data-intensive endeavors. Many such efforts are stymied by privacy concerns and regulatory compliance issues. For example, many hospitals are interested in pooling their medical records for research, but none may disclose arbitrary patient records to researchers or other healthcare providers. In this context we propose the Private Data Network (PDN), a federated database for querying over the collective data of mutually distrustful parties. In a PDN, each member database does not reveal its tuples to its peers nor to the query writer. Instead, the user submits a query to an honest broker that plans and coordinates its execution over multiple private databases using secure multiparty computation (SMC). Here, each database's query execution is oblivious, and its program counters and memory traces are agnostic to the inputs of others.We introduce a framework for executing PDN queries named smcql. This system translates SQL statements into SMC primitives to compute query results over the union of its source databases without revealing sensitive information about individual tuples to peer data providers or the honest broker. Only the honest broker and the querier receive the results of a PDN query. For fast, secure query evaluation, we explore a heuristics-driven optimizer that minimizes the PDN's use of secure computation and partitions its query evaluation into scalable slices.
14 Jun 2017
TL;DR: A machine-checkable denotational semantics for SQL, the de facto language for relational database, for rigorously validating rewrite rules and an automated decision procedure using HoTTSQL for conjunctive queries: a well studied decidable fragment of SQL that encompasses many real-world queries.
Abstract: Every database system contains a query optimizer that performs query rewrites. Unfortunately, developing query optimizers remains a highly challenging task. Part of the challenges comes from the intricacies and rich features of query languages, which makes reasoning about rewrite rules difficult. In this paper, we propose a machine-checkable denotational semantics for SQL, the de facto language for relational database, for rigorously validating rewrite rules. Unlike previously proposed semantics that are either non-mechanized or only cover a small amount of SQL language features, our semantics covers all major features of SQL, including bags, correlated subqueries, aggregation, and indexes. Our mechanized semantics, called HoTT SQL, is based on K-Relations and homotopy type theory, where we denote relations as mathematical functions from tuples to univalent types. We have implemented HoTTSQL in Coq, which takes only fewer than 300 lines of code and have proved a wide range of SQL rewrite rules, including those from database research literature (e.g., magic set rewrites) and real-world query optimizers (e.g., subquery elimination). Several of these rewrite rules have never been previously proven correct. In addition, while query equivalence is generally undecidable, we have implemented an automated decision procedure using HoTTSQL for conjunctive queries: a well studied decidable fragment of SQL that encompasses many real-world queries.
01 Aug 2017
TL;DR: This work proposes R-skyline operators that generalize both skyline and ranking queries by applying the notion of dominance to a set of scoring functions of interest, and discusses the formal properties of these new operators.
Abstract: Traditionally, skyline and ranking queries have been treated separately as alternative ways of discovering interesting data in potentially large datasets. While ranking queries adopt a specific scoring function to rank tuples, skyline queries return the set of non-dominated tuples and are independent of attribute scales and scoring functions. Ranking queries are thus less general, but usually cheaper to compute and widely used in data management systems.We propose a framework to seamlessly integrate these two approaches by introducing the notion of restricted skyline queries (R-skylines). We propose R-skyline operators that generalize both skyline and ranking queries by applying the notion of dominance to a set of scoring functions of interest. Such sets can be characterized, e.g., by imposing constraints on the function's parameters, such as the weights in a linear scoring function. We discuss the formal properties of these new operators, show how to implement them efficiently, and evaluate them on both synthetic and real datasets.
16 Jan 2017
TL;DR: This work shows completeness of the Sparse Orthogonal Vectors problem for the class of first-order properties under fine-grained reductions, the first such completeness result for a standard complexity class.
Abstract: Properties definable in first-order logic are algorithmically interesting for both theoretical and pragmatic reasons. Many of the most studied algorithmic problems, such as Hitting Set and Orthogonal Vectors, are first-order, and the first-order properties naturally arise as relational database queries. A relatively straightforward algorithm for evaluating a property with k+1 quantifiers takes time O(mk) and, assuming the Strong Exponential Time Hypothesis (SETH), some such properties require O(mk−e) time for any e > 0. (Here, >m represents the size of the input structure, i.e., the number of tuples in all relations.)We give algorithms for every first-order property that improves this upper bound to mk/2Θ (S log n), i.e., an improvement by a factor more than any poly-log, but less than the polynomial required to refute SETH. Moreover, we show that further improvement is equivalent to improving algorithms for sparse instances of the well-studied Orthogonal Vectors problem. Surprisingly, both results are obtained by showing completeness of the Sparse Orthogonal Vectors problem for the class of first-order properties under fine-grained reductions. To obtain improved algorithms, we apply the fast Orthogonal Vectors algorithm of References [3, 16].While fine-grained reductions (reductions that closely preserve the conjectured complexities of problems) have been used to relate the hardness of disparate specific problems both within P and beyond, this is the first such completeness result for a standard complexity class.
01 Nov 2017
TL;DR: A new algorithm Hydra is proposed, which overcomes the quadratic runtime complexity in the number of tuples of state-of-the-art DC discovery methods and results in a speedup by orders of magnitude, especially for datasets with a large number of Tuples.
Abstract: Denial constraints (DCs) are a generalization of many other integrity constraints (ICs) widely used in databases, such as key constraints, functional dependencies, or order dependencies. Therefore, they can serve as a unified reasoning framework for all of these ICs and express business rules that cannot be expressed by the more restrictive IC types. The process of formulating DCs by hand is difficult, because it requires not only domain expertise but also database knowledge, and due to DCs' inherent complexity, this process is tedious and error-prone. Hence, an automatic DC discovery is highly desirable: we search for all valid denial constraints in a given database instance. However, due to the large search space, the problem of DC discovery is computationally expensive.We propose a new algorithm Hydra, which overcomes the quadratic runtime complexity in the number of tuples of state-of-the-art DC discovery methods. The new algorithm's experimentally determined runtime grows only linearly in the number of tuples. This results in a speedup by orders of magnitude, especially for datasets with a large number of tuples. Hydra can deliver results in a matter of seconds that to date took hours to compute.
09 May 2017
TL;DR: An approximation algorithm is developed that runs in linearithmic time and guarantees a regret ratio, within any arbitrarily small user-controllable distance from the optimal regret ratio.
Abstract: Finding the maxima of a database based on a user preference, especially when the ranking function is a linear combination of the attributes, has been the subject of recent research. A critical observation is that the em convex hull is the subset of tuples that can be used to find the maxima of any linear function. However, in real world applications the convex hull can be a significant portion of the database, and thus its performance is greatly reduced. Thus, computing a subset limited to $r$ tuples that minimizes the regret ratio (a measure of the user's dissatisfaction with the result from the limited set versus the one from the entire database) is of interest. In this paper, we make several fundamental theoretical as well as practical advances in developing such a compact set. In the case of two dimensional databases, we develop an optimal linearithmic time algorithm by leveraging the ordering of skyline tuples. In the case of higher dimensions, the problem is known to be NPcomplete. As one of our main results of this paper, we develop an approximation algorithm that runs in linearithmic time and guarantees a regret ratio, within any arbitrarily small user-controllable distance from the optimal regret ratio. The comprehensive set of experiments on both synthetic and publicly available real datasets confirm the efficiency, quality of output, and scalability of our proposed algorithms.
Posted Content•
TL;DR: The authors proposed a new semantic program embedding that is learned from program execution traces, where program states expressed as sequential tuples of live variable values not only capture program semantics more precisely, but also offer a more natural fit for Recurrent Neural Networks to model.
Abstract: Neural program embeddings have shown much promise recently for a variety of program analysis tasks, including program synthesis, program repair, fault localization, etc. However, most existing program embeddings are based on syntactic features of programs, such as raw token sequences or abstract syntax trees. Unlike images and text, a program has an unambiguous semantic meaning that can be difficult to capture by only considering its syntax (i.e. syntactically similar pro- grams can exhibit vastly different run-time behavior), which makes syntax-based program embeddings fundamentally limited. This paper proposes a novel semantic program embedding that is learned from program execution traces. Our key insight is that program states expressed as sequential tuples of live variable values not only captures program semantics more precisely, but also offer a more natural fit for Recurrent Neural Networks to model. We evaluate different syntactic and semantic program embeddings on predicting the types of errors that students make in their submissions to an introductory programming class and two exercises on the CodeHunt education platform. Evaluation results show that our new semantic program embedding significantly outperforms the syntactic program embeddings based on token sequences and abstract syntax trees. In addition, we augment a search-based program repair system with the predictions obtained from our se- mantic embedding, and show that search efficiency is also significantly improved.
01 Mar 2017
TL;DR: In this paper, a polynomial mapping to a canonical form of ODs is presented, and a sound and complete set of axioms (inference rules) for canonical ODs are given.
Abstract: Integrity constraints (ICs) are useful for query optimization and for expressing and enforcing application semantics. However, formulating constraints manually requires domain expertise, is prone to human errors, and may be excessively time consuming, especially on large datasets. Hence, proposals for automatic discovery have been made for some classes of ICs, such as functional dependencies (FDs), and recently, order dependencies (ODs). ODs properly subsume FDs, as they can additionally express business rules involving order; e.g., an employee never has a higher salary while paying lower taxes than another employee.We present a new OD discovery algorithm enabled by a novel polynomial mapping to a canonical form of ODs, and a sound and complete set of axioms (inference rules) for canonical ODs. Our algorithm has exponential worst-case time complexity, O(2|R|), in the number of attributes |R| and linear complexity in the number of tuples. We prove that it produces a complete and minimal set of ODs. Using real and synthetic datasets, we experimentally show orders-of-magnitude performance improvements over the prior state-of-the-art.
01 Aug 2017
TL;DR: The need to incorporate aggregation costs in the partitioning model when executing stateful operations in parallel, in order to minimize the overall latency and/or through-put is identified and demonstrated.
Abstract: Stream processing has become the dominant processing model for monitoring and real-time analytics. Modern Parallel Stream Processing Engines (pSPEs) have made it feasible to increase the performance in both monitoring and analytical queries by parallelizing a query's execution and distributing the load on multiple workers. A determining factor for the performance of a pSPE is the partitioning algorithm used to disseminate tuples to workers. Until now, partitioning methods in pSPEs have been similar to the ones used in parallel databases and only recently load-aware algorithms have been employed to improve the effectiveness of parallel execution.We identify and demonstrate the need to incorporate aggregation costs in the partitioning model when executing stateful operations in parallel, in order to minimize the overall latency and/or through-put. Towards this, we propose new stream partitioning algorithms, that consider both tuple imbalance and aggregation cost. We evaluate our proposed algorithms and show that they can achieve up to an order of magnitude better performance, compared to the current state of the art.
09 May 2017
TL;DR: This paper thoroughly investigates the satisfiability and learning problems in a variety of settings, and derives insight on how the different facets of the problem interplay with the size of the database, thereby providing the theoretical foundations necessary for a future implementation of query learning from examples.
Abstract: This paper investigates the problem of reverse engineering, i.e., learning, select-project-join (SPJ) queries from a user-provided example set, containing positive and negative tuples. The goal is then to determine whether there exists a query returning all the positive tuples, but none of the negative tuples, and furthermore, to find such a query, if it exists. These are called the satisfiability and learning problems, respectively. The ability to solve these problems is an important step in simplifying the querying process for non-expert users.This paper thoroughly investigates the satisfiability and learning problems in a variety of settings. In particular, we consider several classes of queries, which allow different combinations of the operators select, project and join. In addition, we compare the complexity of satisfiability and learning, when the query is, or is not, of bounded size. We note that bounded-size queries are of particular interest, as they can be used to avoid over-fitting (i.e., tailoring a query precisely to only the seen examples).In order to fully understand the underlying factors which make satisfiability and learning (in)tractable, we consider different components of the problem, namely, the size of a query to be learned, the size of the schema and the number of examples. We study the complexity of our problems, when considering these as part of the input, as constants or as parameters (i.e., as in parameterized complexity analysis). Depending on the setting, the complexity of satisfiability and learning can vary significantly. Among other results, our analysis also provides new problems that are complete for W[3], for which few natural problems are known. Finally, by considering a variety of settings, we derive insight on how the different facets of our problem interplay with the size of the database, thereby providing the theoretical foundations necessary for a future implementation of query learning from examples.
01 Jul 2017
TL;DR: In this paper, a unified model that integrates convolutional neural network (CNN) with a general pairwise ranking framework, in which three novel ranking loss functions are introduced, is presented to relieve the severe class imbalance problem from NR (not relation).
Abstract: Connections between relations in relation extraction, which we call class ties, are common. In distantly supervised scenario, one entity tuple may have multiple relation facts. Exploiting class ties between relations of one entity tuple will be promising for distantly supervised relation extraction. However, previous models are not effective or ignore to model this property. In this work, to effectively leverage class ties, we propose to make joint relation extraction with a unified model that integrates convolutional neural network (CNN) with a general pairwise ranking framework, in which three novel ranking loss functions are introduced. Additionally, an effective method is presented to relieve the severe class imbalance problem from NR (not relation) for model training. Experiments on a widely used dataset show that leveraging class ties will enhance extraction and demonstrate the effectiveness of our model to learn class ties. Our model outperforms the baselines significantly, achieving state-of-the-art performance.
01 Jan 2017
TL;DR: It is shown that k-RMS is NP-hard even when the dimensionality is 3, which provides a complete characterization of the complexity of k- RMS, and answers an open question in previous studies.
Abstract: We study the k-regret minimizing query (k-RMS), which is a useful operator for supporting multi-criteria decision-making. Given two integers k and r, a k-RMS returns r tuples from the database which minimize the k-regret ratio, defined as one minus the worst ratio between the k-th maximum utility score among all tuples in the database and the maximum utility score of the r tuples returned. A solution set contains only r tuples, enjoying the benefits of both top-k queries and skyline queries. Proposed in 2012, the query has been studied extensively in recent years. In this paper, we advance the theory and the practice of k-RMS in the following aspects. First, we develop efficient algorithms for k-RMS (and its decision version) when the dimensionality is 2. The running time of our algorithms outperforms those of previous ones. Second, we show that k-RMS is NP-hard even when the dimensionality is 3. This provides a complete characterization of the complexity of k-RMS, and answers an open question in previous studies. In addition, we present approximation algorithms for the problem when the dimensionality is 3 or larger.