基于知识增强的文本隐喻识别图编码方法

隐喻识别是自然语言处理中语义理解的重要任务之一,目标为识别某一概念在使用时是否借用了其他概念的属性和特点. 由于单纯的神经网络方法受到数据集规模和标注稀疏性问题的制约,近年来,隐喻识别研究者开始探索如何利用其他任务中的知识和粗粒度句法知识结合神经网络模型,获得更有效的特征向量进行文本序列编码和建模.然而,现有方法忽略了词义项知识和细粒度句法知识,造成了外部知识利用率低的问题,难以建模复杂语境.针对上述问题,提出一种基于知识增强的图编码方法（knowledge-enhanced graph encoding method,KEG）来进行文本中的隐喻识别. 该方法分为3个部分：在文本编码层,利用词义项知识训练语义向量,与预训练模型产生的上下文向量结合,增强语义表示;在图网络层,利用细粒度句法知识构建信息图,进而计算细粒度上下文,结合图循环神经网络进行迭代式状态传递,获得表示词的节点向量和表示句子的全局向量,实现对复杂语境的高效建模;在解码层,按照序列标注架构,采用条件随机场对序列标签进行解码.实验结果表明,该方法的性能在4个国际公开数据集上均获得有效提升.

     

A metalogic programming approach: language,semantics and applications

Abstract

This paper presents a logic programming language of novel conception, called Reflective Prolog, which allows declarative metaknowledge representation and metareasoning. The language is defined by augmenting pure Prolog (Horn clauses) with capabilities of self-reference and logical reflection. Self-reference is designed as a quotation device (a carefully defined naming relation) which allows the construction of metalevel terms that refer to object-level terms and atoms. Logical reflection is designed as an unquotation mechanism (a distinguished truth predicate) which relates names to what is named, thus extending the meaning of domain predicates. The reflection mechanism is embodied in an extended resolution procedure which automatically switches the context between levels. This implicit reflection relieves the programmer from having to explicitly deal with control aspects of the inference process. The declarative semantics of a Reflective Prolog definite program P is provided in terms of the least reflective Herbrand model of P, characterized by means of a suitable mapping defined over the Herbrand interpretations of P. The extended resolution is proved sound and complete with respect to the least reflective Herbrand model. By illustrating Reflective Prolog solutions to an organic set of problems, and by discussing the main differences with respect to other approaches to logic metaprogramming, we show that the proposed language deploys, within its field of action, greater expressive and inferential power than those available till now. The interpreter of the language has been fully implemented. Because of its enhanced power, logic semantics and working interpreter, Reflective Prolog is offered as a contribution toward making the declarative approach of logic programming applicable to the development of increasingly sophisticated knowledge-based systems.     

基于语义的自然语言文本数字水印研究

数字水印技术是信息隐藏技术研究领域的重要分支,也是当今网络信息安全和数字媒体版权保护研究的重要手段之一。目前数字水印技术的研究主要集中在静止图像和视频的保护等方面,文本数字水印研究较少。针对自然语言文本自身的特性,分析和比较了目前主要的文本数字水印方法及其技术特点,提出了文本数字水印的理论目标和攻击模型,给出了一种基于语义的文本数字水印算法,最后展望了文本数字水印技术研究的发展前景。     

自然语言语义理解在反垃圾邮件中的应用

在分析了基于统计的垃圾邮件分类中关键词的语义缺失问题后,提出将基于本体的自然语言语义理解应用于反垃圾邮件中,以解决关键词的语义问题。经过对垃圾邮件的特征进行分析,设计了关键词驱动的句法分析及基于本体的语义推理,在一定程度上解决了邮件过滤过程中词和句子的语义缺失,为垃圾邮件过滤提供了有益的探索。     

Towards a natural language semantics without functors and operands

The paper sets out to offer an alternative to the function/argument approach to the most essential aspects of natural language meanings. That is, we question the assumption that semantic completeness (of, e.g., propositions) or incompleteness (of, e.g., predicates) exactly replicate the corresponding grammatical concepts (of, e.g., sentences and verbs, respectively). We argue that even if one gives up this assumption, it is still possible to keep the compositionality of the semantic interpretation of simple predicate/argument structures. In our opinion, compositionality presupposes that we are able to compare arbitrary meanings in term of information content. This is why our proposal relies on an ‘intrinsically’ type free algebraic semantic theory. The basic entities in our models are neither individuals, nor eventualities, nor their properties, but ‘pieces of evidence’ for believing in the ‘truth’ or ‘existence’ or ‘identity’ of any kind of phenomenon. Our formal language contains a single binary non-associative constructor used for creating structured complex terms representing arbitrary phenomena. We give a finite Hilbert-style axiomatisation and a decision algorithm for the entailment problem of the suggested system.     

ProcessTalk A natural language interface for industrial applications

Many commercial software packages are available for industrial applications. However, small and medium size enterprises, SME, normally lack the in-house qualified personnel to benefit from such packages. As natural languages are the best means of communication, a natural language interface can therefore be a good means to grant the SME an easy access to the sophisticated computerized software packages, thus alleviating the problem of the lack of qualified personnel and reducing the cost of employees' learning curve. At present, no such interfaces exist.

This paper presents the design of a human-machine interface, ProcessTalk, that would accept natural language commands in a text form, then would invoke and execute the appropriate functional/procedural commands required by the desired commercial software package the same way as qualified analysts and engineers would operate. It also exposes its benefits and some of its industrial applications.     

Predicate logic with flexibly binding operators and natural language semantics

A new formalism for predicate logic is introduced, with a non-standard method of binding variables, which allows a compositional formalization of certain anaphoric constructions, including donkey sentences and cross-sentential anaphora. A proof system in natural deduction format is provided, and the formalism is compared with other accounts of this type of anaphora, in particular Dynamic Predicate Logic.Both authors have been supported by grants from the Swedish Council for Research in the Humanities and Social Sciences.     

Algebraic structure of some stochastic discrete event systems,with applications

Generalized semi-Markov processes (GSMPs) and stochastic Petri nets (SPNs) are generally regarded as performance models (as opposed to logical models) of discrete event systems. Here we take the view that GSMPs and SPNS are essentially automata (generators) driven by input sequences that determine the timing of events. This view combines the deterministic, logical aspects and the stochastic, timed aspects of the two models. We focus on two conditions, (M) and (CX) (which we previously developed to study monotonicity and convexity properties of GSMPs), and the antimatroid and lattice structure they imply for the language generated by a GSMP or SPN. We illustrate applications of these structural properties in the areas of derivative estimation, simulation variance reduction, parallel simulation, and optimal control.Research supported in part by NSF grant ECS-89-96230.     

Temporal semantics and natural language processing in a decision support system

This paper is an attempt to address in an integrated fashion three issues of interest in the decision support system (DSS) context. First, to deal with dynamic problem environments, a mechanism is proposed to represent data and decision models with explicit temporal aspects. Second, in order to provide a natural language interface to such a DSS, this representation of temporal knowledge is used to process queries including those with explicit temporal aspects. Third, the issue of computational efficiency has been addressed using the language of equational logic as a deduction oriented language for problem solving as well as describing environmental knowledge. An example describing the salient features of the system is presented.     

Algebraic tools for language processing

A language space provides a unified framework to deal with the properties of language constructs by associating them with their specification rules. The concrete syntax is represented by segments of the language space. The semantics is given by derived operations of the algebras where these constructs are interpreted by the processing tools operating on the language space. We examine in this paper only processing tools that collect syntactic information over the language space. Tools involved in semantics processing such as translators and interpreters are also integrated in the language space but are not discussed here.     

Computability concepts for programming language semantics

This paper is about mathematical problems in programming language semantics and their influence on recursive function theory. We define a notion of computability on continuous higher types (for all types) and show its equivalence to effective operators. This resuit shows that our computable operators can model mathematically (i.e. extensionally) everything that can be done in an operational semantics. These new recursion theoretic concepts which are appropriate to semantics also allow us to construct Scott models for the λ-calculus which contain all and only computably elements. Depending on the choice of the initial cpo, our general theory yields a theory for either strictly determinate or else arbitrary non-deterministic objects (parallelism). The formal theory is developed in Part II of this paper. Part I gives motivation and comparison with related work.     

A Formal Semantics for DAI Language NUML

Traditional AI systems are brittle in the sense that they fail miserably when presented with problems even sliphtly outside of their limited range of expertise.A powerful,extensible strategy of Distributed Artificial Intelligence (DAI) for overcoming such bounds is to put the system in a society of systems.So the ability to coordinate group activities of individuals and to communicate between each other is necessary for a language describing DAI systems.Agent-oriented language NUML is such a language.It is a specific kind of object-oriented language.To give formal semantics to NUML,there is the problem to formalise object-oriented programming paradigm which is still open.The theory of higher-order π-calculus is a concurrent computation model with sufficient capability,which provides us a mathematical tool to do the formalization.This paper tries to use higher-order π-calculus to formalise NUML.     

Execution levels for aspect-oriented programming: Design,semantics, implementations and applications

In aspect-oriented programming (AOP) languages, advice evaluation is usually considered as part of the base program evaluation. This is also the case for certain pointcuts, such as if pointcuts in AspectJ, or simply all pointcuts in higher-order aspect languages like AspectScheme. While viewing aspects as part of base level computation clearly distinguishes AOP from reflection, it also comes at a price: because aspects observe base level computation, evaluating pointcuts and advice at the base level can trigger infinite regression. To avoid these pitfalls, aspect languages propose ad-hoc mechanisms, which increase the complexity for programmers while being insufficient in many cases. After shedding light on the many facets of the issue, this paper proposes to clarify the situation by introducing levels of execution in the programming language, thereby allowing aspects to observe and run at specific, possibly different, levels. We adopt a defensive default that avoids infinite regression, and gives advanced programmers the means to override this default using level-shifting operators. We then study execution levels both in practice and in theory. First, we study the relevance of the issues addressed by execution levels in existing aspect-oriented programs. We then formalize the semantics of execution levels and prove that the default semantics is indeed free of a certain form of infinite regression, which we call aspect loops. Finally, we report on existing implementations of execution levels for aspect-oriented extensions of Scheme, JavaScript and Java, discussing their implementation techniques and current applications.     

CUDL language semantics: Updating FDB data

The semantics for data manipulation of the database language CUDL - Conceptual Universal Database Language - designed to manage dynamic database environments, are presented. This language conforms to the FDB (Frame DataBase) data model, offering a simple, easy and efficient platform for the use of the FDB model. Otherwise the management and operation of FDB data is laborious and time-consuming and it requires from the user a very good acquaintance of the proposed model, the structures and organisation of it as well as the processes of the management of elements that compose it. In this paper we present in depth the semantics of the way of handling the data, in order to search and transform information, in an FDB data source. We present the analysis of simple and complex cases that led us to synthesize valid and simple semantic rules that determine the data manipulation operations. The more sophisticated and demanding constructs, used in the language, for query specification, query processing and object manipulation are discussed and evaluated.     

Linking operational semantics and algebraic semantics for a probabilistic timed shared-variable language

Complex software systems typically involve features like time, concurrency and probability, with probabilistic computations playing an increasing role. However, it is currently challenging to formalize languages incorporating all those features. Recently, the language PTSC has been proposed to integrate probability and time with shared-variable concurrency (Zhu et al. (2006, 2009) [51] and [53]), where the operational semantics has been explored and a set of algebraic laws has been investigated via bisimulation. This paper investigates the link between the operational and algebraic semantics of PTSC, highlighting both its theoretical and practical aspects.The link is obtained by deriving the operational semantics from the algebraic semantics, an approach that may be understood as establishing soundness of the operational semantics with respect to the algebraic semantics. Algebraic laws are provided that suffice to convert any PTSC program into a form consisting of a guarded choice or an internal choice between programs, which are initially deterministic. That form corresponds to a simple execution of the program, so it is used as a basis for an operational semantics. In that way, the operational semantics is derived from the algebraic semantics, with transition rules resulting from the derivation strategy. In fact the derived transition rules and the derivation strategy are shown to be equivalent, which may be understood as establishing completeness of the operational semantics with respect to the algebraic semantics.That theoretical approach to the link is complemented by a practical one, which animates the link using Prolog. The link between the two semantics proceeds via head normal form. Firstly, the generation of head normal form is explored, in particular animating the expansion laws for probabilistic interleaving. Then the derivation of the operational semantics is animated using a strategy that exploits head normal form. The operational semantics is also animated. These animations, which again supports to claim soundness and completeness of the operational semantics with respect to the algebraic, are interesting because they provide a practical demonstration of the theoretical results.     

An axiomatic semantics for the synchronous language Gentzen

We propose an axiomatic semantics for the synchronous language Gentzen, which is an instantiation of the paradigm Timed Concurrent Constraint Programming proposed by Saraswat, Jagadeesan and Gupta. We view Gentzen as a prototype of the class of state-oriented synchronous languages, since it offers the basic constructs that are shared by the languages in the class. Since synchronous concurrency cannot be simulated by arbitrary interleaving, we cannot exploit “head normal forms”, on which axiomatic theories for asynchronous process calculi are based. We suggest how axiomatic semantics for other state-oriented synchronous languages can be obtained by expressing constructs of such languages in terms of Gentzen constructs.     

Translational Semantics for a Conceptual Level Query Language

A conceptual level database language for the entity relationship (ER) model implicitly contains integrities basic to ER concepts and special retrieval semantics for inheritances of attributes and relationships.Prolog,which belongs to the logical and physical level,cannot be used as a foundation to directly define the database language.It is shown how Prolog can be enhanced to understand the concepts of entities,relationships,attributes and is-a relationships.The enhanced Prolog is then used as a foundation to define the semantics of a database query language for the ER model.The three basic functions of model specification,updates and retrievals are defined.     

A layered semantics for a parallel object-oriented language

We develop a denotational semantics for POOL, a parallel object-oriented programming language. The main contribution of this semantics is an accurate mathematical model of the most important concept in object-oriented programming: the object. This is achieved by structuring the semantics in layers working at three different levels: for statements, objects and programs. For each of these levels we define a specialized mathematical domain of processes, which we use to assign a meaning to each language construct. This is done in the mathematical framework of complete metric spaces. We also define operators that translate between these domains. At the program level we give a precise definition of the observable input/output behaviour of a particular program, which could be used at a later stage to decide the issue of full abstractness. We illustrate our semantic techniques by first applying them to a toy language similar to CSP.This paper describes work done in ESPRIT Basic Research Action 3020,Integration.     

Extending the UML concepts to transform natural language queries with fuzzy semantics into SQL

Database applications tend toward getting more versatile and broader to comply with the expansion of various organizations. However, naïve users usually suffer from accessing data arbitrarily by using formal query languages. Therefore, we believe that accessing databases using natural language constructs will become a popular interface in the future. The concept of object-oriented modeling makes the real world to be well represented or expressed in some kinds of logical form. Since the class diagram in UML is used to model the static relationships of databases, in this paper, we intend to study how to extend the UML class diagram representations to capture natural language queries with fuzzy semantics. By referring to the conceptual schema throughout the class diagram representation, we propose a methodology to map natural language constructs into the corresponding class diagram and employ Structured Object Model (SOM) methodology to transform the natural language queries into SQL statements for query executions. Moreover, our approach can handle queries containing vague terms specified in fuzzy modifiers, like ‘good’ or ‘bad’. By our approach, users obtain not only the query answers but also the corresponding degree of vagueness, which can be regarded as the same way we are thinking.     

On open‐set lattices and some of their applications in semantics

In this article, we present the theory of Kripke semantics, along with the mathematical framework and applications of Kripke semantics. We take the Kripke‐Sato approach to define the knowledge operator in relation to Hintikka's possible worlds model, which is an application of the semantics of intuitionistic logic and modal logic. The applications are interesting from the viewpoint of agent interactives and process interaction. We propose (i) an application of possible worlds semantics, which enables the evaluation of the truth value of a conditional sentence without explicitly defining the operator “→” (implication), through clustering on the space of events (worlds) using the notion of neighborhood; and (ii) a semantical approach to treat discrete dynamic process using Kripke‐Beth semantics. Starting from the topological approach, we define the measure‐theoretical machinery, in particular, we adopt the methods developed in stochastic process—mainly the martingale—to our semantics; this involves some Boolean algebraic (BA) manipulations. The clustering on the space of events (worlds), using the notion of neighborhood, enables us to define an accessibility relation that is necessary for the evaluation of the conditional sentence. Our approach is by taking the neighborhood as an open set and looking at topological properties using metric space, in particular, the so‐called ε‐ball; then, we can perform the implication by computing Euclidean distance, whenever we introduce a certain enumerative scheme to transform the semantic objects into mathematical objects. Thus, this method provides an approach to quantify semantic notions. Combining with modal operators K_i operating on E set, it provides a more‐computable way to recognize the “indistinguishability” in some applications, e.g., electronic catalogue. Because semantics used in this context is a local matter, we also propose the application of sheaf theory for passing local information to global information. By looking at Kripke interpretation as a function with values in an open‐set lattice ??_U, which is formed by stepwise verification process, we obtain a topological space structure. Now, using the measure‐theoretical approach by taking the Borel set and Borel function in defining measurable functions, this can be extended to treat the dynamical aspect of processes; from the stochastic process, considered as a family of random variables over a measure space (the probability space triple), we draw two strong parallels between Kripke semantics and stochastic process (mainly martingales): first, the strong affinity of Kripke‐Beth path semantics and time path of the process; and second, the treatment of time as parametrization to the dynamic process using the technique of filtration, adapted process, and progressive process. The technique provides very effective manipulation of BA in the form of random variables and σ‐subalgebra under the cover of measurable functions. This enables us to adopt the computational algorithms obtained for stochastic processes to path semantics. Besides, using the technique of measurable functions, we indeed obtain an intrinsic way to introduce the notion of time sequence. © 2003 Wiley Periodicals, Inc.