支持深度学习的视觉数据库管理系统研究进展

计算机视觉因其强大的学习能力,在各种真实场景中得到了广泛应用.随着数据库的发展,利用数据库中成熟的数据管理技术来处理视觉分析应用,已成为一种日益增长的研究趋势.图像、视频和文本等多模态数据的相互融合处理,也促进了视觉分析应用的多样性和准确性.近年来,因深度学习的兴起,支持深度学习的视觉分析应用开始受到广泛关注.然而,传统的数据库管理技术在深度学习场景下面临着复杂视觉分析语义难以表达、应用执行效率低等问题.因此,支持深度学习的视觉数据库管理系统得到了广泛关注.综述了目前视觉数据库管理系统的研究进展:首先,总结了视觉数据库管理系统在不同层面上面临的挑战,包括编程接口、查询优化、执行调度和数据存储;其次,分别探讨了上述4个层面上的相关技术;最后,对视觉数据库管理系统未来的研究方向进行了展望.     

Schema evolution and integration

Providing support forschema evolution allows existing databases to be adjusted for varying roles over time. This paper reflects on existing evolution support schemes and introduces a more general and functional mechanism to support schema evolution andinstance adaptation for centralized and distributed object-oriented database systems. Our evolution support scheme is distinguished from previous mechanisms in that it is primarily concerned with preserving existing database objects and maintaining compatibility for old applications, while permitting a wider range of evolution operations. It achieves this by supporting schema versioning, allowing multiple representations of instances to persist simultaneously, and providing for programmer specification of how to adapt existing instances. The mechanism is general enough to provide much of the support necessarily forheterogeneous schema integration, as well as incorporating much of the features of object migration and replication.     

A lexical analyser for Arabic

Practical natural language processing (NLP) systems such as database front-ends, deductive databases or object-oriented databases are at the forefront of research into the next-generation intelligent database systems. The research described in this paper has been aimed at integrating front-end paradigms and rule-based deduction to provide a single powerful framework for database systems in Arabic. The lexicon stores only roots of verbs and uses a program intelligent enough to handle all derived forms automatically. This is significant, as these alone represent 70% of the total dictionary. As part of the discussion of this system, its utility in such NLP applications as parsing and machine translation is examined.     

Design and Implementation of the ROL Deductive Object-Oriented Database System

ROL is a deductive object-oriented database system that has been implemented at the University of Regina. It provides a uniform rule-based declarative language for defining, manipulating and querying databases, which integrates important features of both deductive databases and object-oriented databases. It supports object identity, complex objects, classes, class hierarchies, multiple inheritance with overriding and blocking, and schema definition. It also supports structured values such as functor objects and sets, treating them as first class citizens, and providing powerful mechanisms for representing both partial and complete information about sets. This paper describes its design and implementation. An important novel feature of the implementation is that it combines top-down and bottom-up evaluation strategies and automatically selects a strategy based on the nature of the query and data in the database.     

Modularizing application and database evolution – an aspect‐oriented framework for orthogonal persistence

In the maintenance of software applications, database evolution is one common difficulty. In object‐oriented databases, this process comprises schema evolution and instance adaptation. Both tasks usually require significant effort from programmers and database administrators. In this paper, we propose orthogonal persistence and aspect‐oriented programming to support semi‐transparent database evolution. A default mechanism for instance evolution is defined, but the user may provide modularized solutions using the aspect‐oriented paradigm. We present our framework AOF4OOP to test the feasibility of our proposed approach. This prototype allows programmes to transparently access data in other versions of the database schema. We evaluate our framework, comparing it to related approaches using two real applications and measuring the improvement of the productivity of the programmer. Copyright © 2016 John Wiley & Sons, Ltd.     

Inductive database languages: requirements and examples

Inductive databases (IDBs) represent a database perspective on Knowledge discovery in databases (KDD). In an IDB, the KDD application can express both queries capable of accessing and manipulating data, and queries capable of generating, manipulating, and applying patterns allowing to formalize the notion of mining process. The feature that makes them different from other data mining applications is exactly the idea of looking at the support for knowledge discovery as an extension of the query process. This paper draws a list of desirable properties to be taken into account in the definition of an IDB framework. They involve several dimensions, such as the expressiveness of the language in representing data and models, the closure principle, the capability to provide a support for an efficient algorithm programming. These requirements are a basis for a comparative study that highlights strengths and weaknesses of existing IDB approaches. The paper focuses on the SQL-based ATLaS language/system, on the logic-based LDL++{\mathcal{LDL}++} language/system, and on the XML-based KDDML language/system.     

A logic-based integration of active and deductive databases

A logic-based approach to the specification of active database functionality is presented which not only endows active databases with a well-defined and well-understood formal semantics, but also tightly integrates them with deductive databases. The problem of endowing deductive databases with rule-based active behaviour has been addressed in different ways. Typical approaches include accounting for active behaviour by extending the operational semantics of deductive databases, or, conversely, accounting for deductive capabilities by constraining the operational semantics of active databases. The main contribution of the paper is an alternative approach in which a class of active databases is defined whose operational semantics is naturally integrated with the operational semantics of deductive databases without either of them strictly subsuming the other. The approach is demonstrated via the formalization of the syntax and semantics of an active-rule language that can be smoothly incorporated into existing deductive databases, due to the fact that the standard formalization of deductive databases is reused, rather than altered or extended. One distinctive feature of the paper is its use of ahistory, as defined in the Kowalski-Sergot event-calculus, to define event occurrences, database states and actions on these. This has proved to be a suitable foundation for a comprehensive logical account of the concept set underpinning active databases. The paper thus contributes a logical perspective to the ongoing task of developing a formal theory of active databases. Alvaro Adolfo Antunes Fernandes, Ph.D.: He received a B.Sc. in Economics (Rio de Janeiro, 1984), an M.Sc. in Knowledge-Based Systems (Edinburgh, 1990) and a Ph.D. in Computer Science (Heriot-Watt, 1995). He worked as a Research Associate at Heriot-Watt University from December 1990 until December 1995. In January 1996 he joined the Department of Mathematical and Computing Sciences at Goldsmiths College, University of London, as a Lecturer. His current research interests include advanced data- and knowledge-base technology, logic programming, and software engineering. M. Howard Williams, Ph.D., D.Sc.: He obtained his Ph.D. in ionospheric physics and recently a D.Sc. in Computer Science. He was appointed as the first lecturer in Computer Science at Rhodes University in 1970. During the following decade he rose to Professor of Computer Science and in 1980 was appointed as Professor of Computer Science at Heriot-Watt University. From 1980 to 1988 he served as Head of Department and then as director of research until 1992. He is now head of the Database Research Group at Heriot-Watt University. His current research interests include active databases, deductive objectoriented databases, spatial databases, parallel databases and telemedicine. Norman W. Paton, Ph.D.: He received a B.Sc. in Computing Science from the University of Aberdeen in 1986. From 1986 to 1989 he worked as a Research Assistant at the University of Aberdeen, receiving a Ph. D. in 1989. From 1989 to 1995 he was a Lecturer in Computer Science at Heriot-Watt University. Since July 1995, he has been a Senior Lecturer in Department of Computer Science at the University of Manchester. His current research interests include active databases, deductive object-oriented databases, spatial databases and database interfaces.     

Object-oriented data modeling in rule-based software development environments

Object-oriented databases (OODBs) were developed to support advanced applications for which traditional databases are not sufficient. The data management requirements of these new applications are significantly different from more traditional data processing applications. More light needs to be shed on these requirements in order to identify the aspects of OODBs that can lead to standards. We have studied the data management requirements of one class of advanced database applications: rule-based software development environments (RBDEs). RBDEs store project components in an object database and control access to these objects through a rule-based process model, which defines each development activity as a rule. The components are abstracted as instances of classes which are defined by the project's data model. In this paper we discuss the constructs that a data modeling language for RBDEs should provide, and then explore some of the data management requirements of RBDEs. We use the Marvel system we developed at Columbia as an example.     

A Hybrid Fragmentation Approach for Distributed Deductive Database Systems

Fragmentation of base relations in distributed database management systems increases the level of concurrency and therefore system throughput for query processing. Algorithms for horizontal and vertical fragmentation of relations in relational, object-oriented and deductive databases exist; however, hybrid fragmentation techniques based on variable bindings appearing in user queries and query-access-rule dependency are lacking for deductive database systems. In this paper, we propose a hybrid fragmentation approach for distributed deductive database systems. Our approach first considers the horizontal partition of base relations according to the bindings imposed on user queries, and then generates vertical fragments of the horizontally partitioned relations and clusters rules using affinity of attributes and access frequency of queries and rules. The proposed fragmentation technique facilitates the design of distributed deductive database systems. Received 4 August 1999 / Revised 30 March 2000 / Accepted in revised form 6 October 2000     

Extending Datalog with Declarative Updates

The semantics of static deductive databases is well understood based on the work in logic programming. In the past decade, various methods to incorporate update constructs into logic programming and deductive databases have been proposed. However, there is still no consensus about the appropriate treatment of dynamic behavior in deductive databases. In this paper, we propose a language called DatalogU, which is a minimal but powerful extension of Datalog with updates to base relations. DatalogU allows the user to program set-oriented complex database transactions with concurrent, disjunctive and sequential update operations in a simple and direct way. It has a simple and intuitive declarative semantics that naturally accounts for set-oriented updates in deductive databases.     

Perspectives in deductive databases

I discuss my experiences, some of the work that I have done, and related work that influenced me, concerning deductive databases, over the last 30 years. I divide this time period into three roughly equal parts: 1957–1968, 1969–1978, 1979–present. For the first I describe how my interest started in deductive databases in 1957, at a time when the field of databases did not even exist. I describe work in the beginning years, leading to the start of deductive databases about 1968 with the work of Cordell Green and Bertram Raphael. The second period saw a great deal of work in theorem providing as well as the introduction of logic programming. The existence and importance of deductive databases as a formal and viable discipline received its impetus at a workshop held in Toulouse, France, in 1977, which culminated in the book Logic and Data Bases. The relationship of deductive databases and logic programming was recognized at that time. During the third period we have seen formal theories of databases come about as an outgrowth of that work, and the recognition that artificial intelligence and deductive databases are closely related, at least through the so-called expert database systems. I expect that the relationships between techniques from formal logic, databases, logic programming, and artificial intelligence will continue to be explored and the field of deductive databases will become a more prominent area of computer science in coming years.     

Coping with indefinite and negative data in deductive databases: A survey

A problem with current database systems is the limitation placed on the type of data which may be represented and manipulated within such systems. In an attempt to broaden this to a wider class of data (i.e. rules as well as facts) and a more powerful set of manipulations, the concept of a deductive database was introduced. However, for the sake of efficiency the type of rule which is allowed in a deductive database is restricted in form. This paper surveys a number of attempts to move towards less restrictive forms of rules in deductive databases which allow indefinite and negative data to be handled.     

Overview of emerging database architectures

A database system is a miniworld data repository representing some aspects of the real world. In a database system, information is well organized and has some inherent meaning. In recent years, the emergence of new types of computer applications, such as multimedia and computer-added manufacturing, have resulted in demands to manage complex data types that are beyond the capabilities of existing relational databases. At the same time, new developments in network technologies, artificial intelligence, semantic modeling, and object-oriented programming have inspired research in database topics such as distributed, deductive, object-oriented, active, and deductive and object-oriented databases (DOOD). Essentially, the most promising new database technologies combine the merits of different kinds of database architectures. Starting with the relational database architecture, this paper will review the emerging database technologies, and thereby, provide insight into current and future directions in database technologies.     

Relaxation as a platform for cooperative answering

Responses to queries posed by a user of a database do not always contain the information desired. Database answers to a query, although they may be logically correct, can sometimes be misleading. Research in the area of cooperative answering for databases and deductive databases seeks to rectify these problems. We introduce a cooperative method calledrelaxation for expanding deductive database and logic programming queries. The relaxation method expands the scope of a query by relaxing the constraints implicit in the query. This allows the database to return answers related to the original query as well as the literal answers themselves. These additional answers may be of interest to the user. In section 1 we introduce the problem and method. In Section 2 we give some background on the research done in cooperative answering. Section 3 discusses the relaxation method, a potential control strategy, and uses. Section 4 looks at a semantic counterpart to this notion. In Section 5 we explore some of the control and efficiency issues. We enumerate open issues in Section 6, and conclude in Section 7.     

Classifying and detecting anomalies in hybrid knowledge-based systems

The increasing need for hybrid Knowledge-Based Systems (KBS) that accommodate more complex applications has led to the need for new verification concerns that are more specific to the hybrid representation using objects and rule-based inference. Traditionally, verification of expert systems has focused solely on rule-based inference systems. Hybrid KBSs present additional verification problems not found in traditional rule-based systems. This paper is an investigation into the anomalies that may be present in a hybrid representation that warrant detection for the verification of the KBS. Many anomalies are due to the interaction of the component parts of the hybrid KBS. For example, subsumption anomalies arise due to an interaction between inheritance of objects and rule-based inference. In this paper, we extend the context of subsumption anomalies and introduce additional types of anomalies that may be present in the KBS. The goal of this research is to make hybrid KBSs more reliable by detecting such anomalies.     

BilVideo: Design and Implementation of a Video Database Management System

With the advances in information technology, the amount of multimedia data captured, produced, and stored is increasing rapidly. As a consequence, multimedia content is widely used for many applications in today’s world, and hence, a need for organizing this data, and accessing it from repositories with vast amount of information has been a driving stimulus both commercially and academically. In compliance with this inevitable trend, first image and especially later video database management systems have attracted a great deal of attention, since traditional database systems are designed to deal with alphanumeric information only, thereby not being suitable for multimedia data.In this paper, a prototype video database management system, which we call BilVideo, is introduced. The system architecture of BilVideo is original in that it provides full support for spatio-temporal queries that contain any combination of spatial, temporal, object-appearance, external-predicate, trajectory-projection, and similarity-based object-trajectory conditions by a rule-based system built on a knowledge-base, while utilizing an object-relational database to respond to semantic (keyword, event/activity, and category-based), color, shape, and texture queries. The parts of BilVideo (Fact-Extractor, Video-Annotator, its Web-based visual query interface, and its SQL-like textual query language) are presented, as well. Moreover, our query processing strategy is also briefly explained.This work is partially supported by the Scientific and Research Council of Turkey (TÜBİTAK) under Project Code 199E025, Turkish State Planning Organization (DPT) under Grant No. 2004K120720, and European Union under Grant No. FP6-507752 (MUSCLE Network of Excellence Project).     

A frame-based model for large manufacturing databases

A major requirement of database systems in engineering design and manufacturing applications is support for storage and maintenance of complex objects. Frame based systems are capable of modeling complex objects. However, many of these systems are implemented in main memory. As the number of objects to be stored far exceeds the capacity of the main memory of a computer, such an implementation is often unusable. Therefore, new ways to model and manipulate large manufacturing databases are needed. In this paper we present an implementation of a frame based system on top of the POSTGRES extensible database system. Such an implementation combines the advantages of database management and frame based systems and allows for the development of large manufacturing database applications with minimal effort.     

Inductive learning in deductive databases

Most current applications of inductive learning in databases take place in the context of a single extensional relation. The authors place inductive learning in the context of a set of relations defined either extensionally or intentionally in the framework of deductive databases. LINUS, an inductive logic programming system that induces virtual relations from example positive and negative tuples and already defined relations in a deductive database, is presented. Based on the idea of transforming the problem of learning relations to attribute-value form, several attribute-value learning systems are incorporated. As the latter handle noisy data successfully, LINUS is able to learn relations from real-life noisy databases. The use of LINUS for learning virtual relations is illustrated, and a study of its performance on noisy data is presented     

Programming with non-determinism in deductive databases

While non-determinism has long been established as a key concept in logic pro-gramming, its importance in the context of deductive databases was recognized only recently. This paper provides an overview of recent results on this topic with the aim of providing an introduction to the theory and practice of non-determinism in deductive databases. In particular we (i) recall the main results linking non-deterministic constructs in database languages to the theory of data complexity and the expressibility hierarchy of query languages; (ii) provide a reasoned introduction to effective programming with non-deterministic constructs; (iii) compare the usage of non-deterministic constructs in languages such as LDL++ to that of traditional logic programming languages; (iv) discuss the link between the semantics of logic programs with non-deterministic constructs and the stable-model semantics of logic programs with negation.