Kewen Wang

Griffith University · School of Information and Communication Technology (ICT)

Visual Question Answering (VQA) systems are known for their poor performance in out-of-distribution datasets. An issue that was addressed in previous works through ensemble learning, answer re-ranking, or artificially growing the training set. In this work, we show for the first time that robust Visual Question Answering is attainable by simply enh...

Multiple-choice question answering (MCQA) becomes particularly challenging when all choices are relevant to the question and are semantically similar. Yet this setting of MCQA can potentially provide valuable clues for choosing the right answer. Existing models often rank each choice separately, overlooking the context provided by other choices. Sp...

Multiple-choice question answering (MCQA) becomes particularly challenging when all choices are relevant to the question and are semantically similar. Yet this setting of MCQA can potentially provide valuable clues for choosing the right answer. Existing models often rank each choice separately, overlooking the context provided by other choices. Sp...

Rule learning is a machine learning method that extracts implicit rules and patterns from data, enabling symbol-based reasoning in artificial intelligence. Unlike data-driven approaches such as deep learning, using rules for inference allows for interpretability. Many studies have attempted to automatically learn first-order rules from knowledge gr...

Designing incentive-compatible and revenue-maximizing auctions is pivotal in mechanism design. Often referred to as optimal auction design, the area has seen little theoretical breakthrough since Myerson’s 1981 seminal work. Not to mention general combinatorial auctions, we don’t even know the optimal auction for selling as few as two distinct item...

Knowledge graphs are often highly incomplete due to their large sizes and one major task for knowledge graph completion is entity typing, that is to predict missing types of entities or vice versa. It is especially challenging to perform entity typing when the type is new, i.e., unseen during training, which is known as the zero-shot entity typing...

Texts contain a vast amount of useful information. Converting unstructured text into structured data enables machines to better understand and mine useful information within texts. Knowledge graphs provide a natural method for data representation and can be applied in downstream tasks. However, existing knowledge graph construction (KGC) methods ar...

Relation classification aims to recognize semantic relation between two given entities mentioned in the given text. Existing models have performed well on the inverse relation classification with large-scale datasets, but their performance drops significantly for few-shot learning. In this paper, we propose a Phrase-level Attention Network, functio...

The task of question answering is to find the most appropriate answer for an input question in natural language from a given custom knowledge base of information. While the performance of question answering systems has been significantly improved, they still struggle to answer questions that require commonsense reasoning. To capture common sense be...

Relation classification is to recognize semantic relation between two given entities mentioned in the given text in Knowledge Graph. Existing models have performed well on the inverse relation classification with large-scale datasets, but their performance drops significantly for few-shot learning. In this paper, we propose a novel method, function...

The relation classification is to identify semantic relations between two entities in a given text. While existing models perform well for classifying inverse relations with large datasets, their performance is significantly reduced for few-shot learning. In this paper, we propose a function words adaptively enhanced attention framework (FAEA) for...

DeGroot-style opinion formation presumes a continuous interaction among agents of a social network. Hence, it cannot handle agents external to the social network that interact only temporarily with the permanent ones. Many real-world organisations and individuals fall into such a category. For instance, a company tries to persuade as many as possib...

The relation classification is to identify semantic relations between two entities in a given text. While existing models perform well for classifying inverse relations with large datasets, their performance is significantly reduced for few-shot learning. In this paper, we propose a function words adaptively enhanced attention framework (FAEA) for...

Dependency-based models for the named entity recognition (NER) task have shown promising results by capturing long-distance relationships between words in a sentence. However, while existing models focus on the syntactic dependency between entities, we are unaware of any work that considers semantic dependency. In this work, we study the usefulness...

Tuple-generating dependencies (TGDs or existential rules) are an expressive constraint language for ontology-mediated query answering and thus query answering is of high complexity. Existing systems based on first-order rewriting methods can lead to queries too large for DBMS to handle. It is shown that datalog rewriting can result in more compact...

Existential rules are an expressive ontology formalism for ontology-mediated query answering and thus query answering is of high complexity, while several tractable fragments have been identified. Existing systems based on first-order rewriting methods can lead to queries too large for DBMS to handle. It is shown that datalog rewriting can result i...

It is natural and effective to use rules for representing explicit knowledge in knowledge graphs. However, it is challenging to learn rules automatically from very large knowledge graphs such as Freebase and YAGO. This paper presents a new approach, RLvLR (Rule Learning via Learning Representations), to learning rules from large knowledge graphs by...

Mining logical rules from knowledge graphs (KGs) is an important yet challenging task, especially when the relevant data is sparse. Transfer learning is an actively researched area to address the data sparsity issue, where a predictive model is learned for the target domain from that of a similar source domain. In this paper, we propose a novel met...

AGM contraction and revision assume an underlying logic that contains propositional logic. Consequently, this assumption excludes many useful logics such as the Horn fragment of propositional logic and most description logics. Our goal in this paper is to generalise AGM contraction and revision to (near-)arbitrary fragments of classical first-order...

Contemporary approaches for the Semantic Web include hybrid knowledge bases that combine ontologies with rule-based languages. Despite a number of existing combination approaches, little attention has been given to change mechanisms for hybrid knowledge bases that can appropriately handle the dynamics of information on the Web. We present here thre...

Recent methods have adapted the well-established AGM and belief base frameworks for belief change to cover belief revision in logic programs. In this study here, we present two new sets of belief change operators for logic programs. They focus on preserving the explicit relationships expressed in the rules of a program, a feature that is missing in...

Recent methods have adapted the well-established AGM and belief base frameworks for belief change to cover belief revision in logic programs. In this study here, we present two new sets of belief change operators for logic programs. They focus on preserving the explicit relationships expressed in the rules of a program, a feature that is missing in...

Possibilistic logic is a weighted logic for dealing with incomplete and uncertain information by assigning weights to propositional formulas. A possibilistic knowledge base (KB) is a finite set of such formulas. The problem of revising a possibilistic KB by possibilistic formula is not new. However, existing approaches are limited in two ways. Firs...

The critical behaviors of NP-complete problems have been studied extensively, and numerous results have been obtained for Boolean formula satisfiability (SAT) and constraint satisfaction (CSP), among others. However, few results are known for the critical behaviors of NP-hard nonmonotonic reasoning problems so far and in particular, a mathematical...

Two essential tasks in managing description logic knowledge bases are eliminating problematic axioms and incorporating newly formed ones. Such elimination and incorporation are formalised as the operations of contraction and revision in belief change. In this paper, we deal with contraction and revision for the DL-Lite family through a model-theore...

Answer set programming (ASP) has been extended to possibilistic ASP (PASP), in which the notion of possibilistic stable models is defined for possibilistic logic programs. However, possibilistic inferences that correspond to the three inferences in ordinary possibilistic logic have not been explored in PASP yet. In this paper, based on the skeptica...

Ontology-mediated data access and management systems are rapidly emerging. Besides standard query answering, there is also a need for such systems to be coupled with explanation facilities, in particular to explain missing query answers (i.e. desired answers of a query which are not derivable from the given ontology and data). This support is highl...

Ontology engineering and maintenance require (semi-)automated ontology change operations. Intensive research has been conducted on TBox and ABox changes in description logics (DLs), and various change operators have been proposed in the literature. Existing operators largely fall into two categories: syntaxbased and model-based.While each approach...

Multi-context systems (MCS) presented by Brewka and Eiter can be considered as a promising way to interlink decentralized and heterogeneous knowledge contexts. In this paper, we propose preferential multi-context systems (PMCS), which provide a framework for incorporating a total preorder relation over contexts in a multi-context system. In a given...

With the current upward trend in semantically annotated data, ontology-based data access (OBDA) was formulated to tackle the problem of data integration and query answering, where an ontology is formalized as a description logic TBox. In order to meet usability requirements set by users, efforts have been made to equip OBDA system with explanation...

This paper proposes a model, the linear model, for randomly generating logic programs with low density of rules and investigates statistical properties of such random logic programs. It is mathematically shown that the average number of answer sets for a random program converges to a constant when the number of atoms approaches infinity. Several ex...

Two essential tasks in managing Description Logic (DL) ontologies are eliminating problematic axioms and incorporating newly formed axioms. Such elimination and incorporation are formalised as the operations of contraction and revision in belief change.In this paper, we deal with contraction and revision for the DL-Lite family through a model-theor...

Forgetting is an important tool for reducing ontologies by eliminating some redundant concepts and roles while preserving sound and complete reasoning. Attempts have previously been made to address the problem of forgetting in relatively simple description logics (DLs), such as DL-Lite and extended . However, the issue of forgetting for ontologies...

In this paper we investigate forgetting in disjunctive logic programs, where forgetting an atom from a program amounts to a reduction in the signature of that program. The goal is to provide an approach that is syntax-independent, in that if two programs are strongly equivalent, then the results of forgetting an atom in each program should also be...

This paper presents an automata-based algorithm for answering the \emph{provenance-aware} regular path queries (RPQs) over RDF graphs on the Semantic Web. The provenance-aware RPQs can explain why pairs of nodes in the classical semantics appear in the result of an RPQ. We implement a parallel version of the automata-based algorithm using the Prege...

In this paper, we develop a notion of forgetting for normal logic programs under the well-founded semantics. We show that a number of desirable properties are satisfied by our approach. Three different algorithms are presented that maintain the computational complexity of the well-founded semantics, while partly keeping its syntactical structure.

Brewka and Eiter’s nonmonotonic multi-context system is an elegant knowledge representation framework to model heterogeneous and nonmonotonic multiple contexts. Belief change is a central problem in knowledge representation and reasoning. In this paper we follow the classical AGM approach to investigate belief change in multi-context systems. Speci...

A new semantic forgetting for answer set programs (ASP), called SM-forgetting, is proposed in the paper. It distinguishes itself from the others in that it preserves not only skeptical and credulous consequences on unforgotten variables, but also strong equivalence-forgetting same variables in strongly equivalent logic programs has strongly equival...

Although several proposals to combine description logics with logic programming rules have been brought forward, hardly any of these approaches capture the dynamic nature of the Semantic Web. In this paper, we look at an expressive combination formalism, normal DL logic programs, and address changes to the rule component from the viewpoint of belie...

DL-Lite is an important family of description logics. Recently, there is an increasing interest in handling inconsistency in DL-Lite as the constraint imposed by a TBox can be easily violated by assertions in ABox in DL-Lite. In this paper, we present a distance-based paraconsistent semantics based on the notion of feature in DL-Lite, which provide...

The answer set semantics presented by Faber et al. [27] has been widely used to define so called FLP answer sets for different types of logic programs. However, it was recently observed that when being extended from normal to more general classes of logic programs, this approach may produce answer sets with circular justifications that are caused b...

A number of proposals have been proposed for measuring inconsistency for knowledge bases. However, it is rarely investigated how to incorporate preference information into inconsistency measures. This paper presents two approaches to measuring inconsistency for stratified knowledge bases. The first approach, termed the multi-section inconsistency m...

We study the spelling suggestion problem for keyword search on XML documents. To address the problems in existing work, we propose a distance-based approach to suggesting meaningful query candidates for an issued query. Our approach uses distance to measure the relationship between keyword matching nodes, and ranks a candidate higher if there are c...

This paper makes the first attempt to establish a framework for possibilistic reasoning in (nonmonotonic) multi-context systems, called possibilistic MCS. We first introduce the syntax for possibilistic MCS and then define its equilibrium semantics based on Brewka and Eiter's nonmonotonic multi-context systems. Then we investigate several propertie...

Ontology construction in OWL is an important and yet time-consuming task even for knowledge engineers and thus a (semi-) automatic approach will greatly assist in constructing ontologies. In this paper, we propose a novel approach to learning concept definitions in $\ensuremath{\ensuremath{\cal E}\ensuremath{\cal L}^{++}} $ from a collection of ass...

The problem of extending description logics with uncertainty has received significant attention in recent years. In this paper, we investigate a probabilistic extension of DL-Lite, a family of tractable description logics. We first present a new probabilistic semantics for terminological knowledge bases based on the notion of types. The semantics p...

Ontologies have been widely used in advanced information systems. How-ever, it has been a challenging issue in ontology engineering to efficiently revise ontolo-gies as new information becomes available. A novel method of revising ontologies has been proposed recently by Wang et al. However, related algorithms have not been im-plemented yet. In thi...

The concept of forgetting has received significant interest in artificial intelligence recently. Informally, given a knowledge base, we may wish to forget about (or discard) some redundant parts (such as atoms, predicates, concepts, etc) but still preserve the consequences for certain forms of reasoning. In nonmonotonic reasoning, so far forgetting...

Logic programming under the stable model semantics has been extended to arbitrary formulas. A question of interest is how to characterize the property of well-supportedness, in the sense of Fages, which has been considered a cornerstone in answer set programming. In this paper, we address this issue by considering general logic programs, which cons...

The FLP semantics presented by (Faber, Leone, and Pfeifer 2004) has been widely used to define answer sets, called FLP answer sets, for different types of logic programs such as logic programs with aggregates, de-scription logic programs (dl-programs), Hex programs, and logic programs with first-order formulas (general logic programs). However, it...

The concept of contexts is widely used in artificial intelligence. Several recent attempts have been made to formalize multi-context systems (MCS) for ontology applications. However, these approaches are unable to handle probabilistic knowledge. This paper introduces a formal framework for representing and reasoning about uncertainty in multi-conte...

Recently much attention has been directed to extending logic programming with description logic (DL) expressions, so that logic programs have access to DL knowledge bases and thus are able to reason with ontologies in the Semantic Web. In this paper, we propose a new extension of logic programs with DL expressions, called normal DL logic programs....

This paper proposes a paraconsistent and nonmonotonic extension of description logic by planting a nonmonotonic mechanism called minimal inconsistency in paradoxical description logics, which is a paraconsistent version of description logics. A precedence relation between two paradoxical models of knowledge bases is firstly introduced to obtain min...

Revising knowledge bases (KBs) in description logics (DLs) in a syntax-independent manner is an important and nontrivial problem for ontology management and DL communities. Several attempts have been made to adapt classical modelbased belief revision/update techniques to DLs, but they are restricted in several ways. In particular, they rarely inves...

In this paper, we propose two new approaches to forgetting for [Ascr ][Lscr ][Cscr ] based on the well-known tableau algorithm. The first approach computes the result of forgetting by rolling up tableaux, and also provides a decision algorithm for the existence of forgetting in [Ascr ][Lscr ][Cscr ]. When the result of forgetting does not exist, we...

Revising knowledge bases (KBs) in description logics (DLs) in a syntax-independent manner is an important, nontrivial problem for the ontology management and DL communities. Several attempts have been made to adapt classical modelbased belief revision and update techniques to DLs, but they are restricted in several ways. In particular, they do not...

Revising knowledge bases (KBs) in description logics (DLs) in a syntax-independent manner is an important, nontrivial problem for the ontology management and DL communities. Several attempts have been made to adapt classical model-based belief revision and update techniques to DLs, but they are restricted in several ways. In particular, they do not...

To support the reuse and combination of ontologies in Semantic Web applications, it is often necessary to obtain smaller ontologies from existing larger ontologies. In particular, applications may require the omission of certain terms, e. g., concept names and role names, from an ontology. However, the task of omitting terms from an ontology is cha...

As a vision for the future of the Web, the Semantic Web is an open, constantly changing and collaborative environment. Hence it is reasonable to expect that knowledge sources in the Semantic Web contain noise and inaccuracies. However, as the logical foundation of Ontology Web Language in the Semantic Web, description logics fail to tolerate incons...

The notion of uniform interpolation for description logic ALC\mathcal{ALC} has been introduced in [9]. In this paper, we reformulate the uniform interpolation for ALC\mathcal{ALC} from the angle of forgetting and show that it satisfies all desired properties of forgetting. Then we introduce an algorithm for computing the result of forgetting in con...

Forgetting is an important tool for reducing ontologies by eliminating some concepts and roles while preserving sound and complete reasoning. Attempts have previously been made to address the problem of forgetting in relatively simple description logics (DLs) such as DL-Lite and extended \({\mathcal {EL}}\). The ontologies used in these attempts we...

Contained rewriting and maximal contained rewriting of tree pattern queries using views have been studied recently for the class of tree patterns involving /, //, and []. Given query Q and view V, it has been shown that a contained rewriting of Q using V can be obtained by finding a useful embedding of Q in V. However, for the same Q and V, there m...

We address the revision problem for knowledge bases (KBs) in Description Logics (DLs). This problem has received much attention in the ontology management and DL communities, but the existing proposals are restricted in several ways. In this paper we develop a formal framework for revision of DL-Lite KBs, using techniques that are analogous to thos...

The notion of forgetting, also known as variable elimination, has been investigated extensively in the context of classical logic, but less so in (nonmonotonic) logic programming and nonmonotonic reasoning. The few approaches that exist are based on syntactic modifications of a program at hand. In this paper, we establish a declarative theory of fo...

To support the reuse and combination of ontologies in Semantic Web applications, it is often necessary to obtain smaller ontologies from existing larger ontologies. In particular, applications may require the omission of many terms, e.g., concept names and role names, from an ontology. However, the task of omit- ting terms from an ontology is chall...

Default logic is an important method of knowledge representation and reasoning, because it supports reasoning with incomplete information, and because defaults can be found naturally in many application domains, such as diagnostic problems, information retrieval, legal reasoning, regulations, specifications of systems and software, etc. Default log...

The language of HEX-programs under the answer-set semantics is designed for interoperating with heterogeneous sources via external atoms and for meta-reasoning via higher-order literals in the context of the semantic Web. As an important technique in managing knowledge bases, the notion of forgetting has received increasing interest in the knowledg...

Several proposals have been put forward to support distributed agent cooperation in the Semantic Web, by allowing concepts and roles in one ontology be reused in another ontology. In general, these proposals reduce the autonomy of each ontology by defining the semantics of the ontology to depend on the semantics of the other ontologies. We propose...

A novel declarative approach of forgetting in answer set programming (ASP) has been proposed recently. In this paper we report a system prototype of forgetting in ASP, called LPForget. It consists of two modules: (1) Forgetting: computing the result of forgetting about certain literals in logic program under the answer set semantics; (2) CRS: appli...

We establish a declarative theory of forgetting for disjunctive logic programs. The suitability of this theory is justified by a number of desirable properties. In particular, one of our results shows that our notion of forgetting is completely captured by the classical forgetting. A transformation-based algorithm is also developed for computing th...

Intentional agents must be aware of their success and failure to truly assess their own progress towards their intended goals. However, our analysis of intentional agent systems indicate that existing architectures are inadequate in this regard. Specifically, existing systems provide few, if any, mechanisms for monitoring for the failure of behavio...

The language of dl-programs is a latest effort in developing an expressive representation for Web-based ontologies. It allows to build answer set programming (ASP) on top of description logic and thus some attractive features of ASP can be employed in the design of the Semantic Web architecture. In this paper we first generalize dl-programs by allo...

Rational agents must be aware of their success and failure to truly assess their own progress towards their intended goals. In this study we describe a detailed investigation of how current BDI agents monitor their successes and failures during their reasoning cycle. Our analysis indicates that the existing architectures are inadequate to specifica...

Nested logic programs and epistemic logic programs are two important extensions of answer set programming. However, the relationship between these two formalisms is rarely explored. In this paper we first introduce the epistemic HT-logic, and then propose a more general extension of logic programs called {\em nested epistemic logic programs}. The s...

We consider how to forget a set of atoms in a logic program. Intuitively, when a set of atoms is forgot- ten from a logic program, all atoms in the set should be eliminated from this program in some way, and other atoms related to them in the program might also be affected. We dene notions of strong and weak forgettings in logic programs to capture...

The study of forgetting for reasoning has attracted considerable attention in AI. However, much of the work on forgetting, and other related approaches such as independence, irrelevance and novelty, has been restricted to the classical logics. This paper describes a detailed theoretical investigation of the notion of forgetting in the context of lo...

For intentional agents to be rational they must be aware of their success and failure to truly assess their own progress towards their intended goals. In this study we describe a detailed investigation of how current intentional agents monitor their successes and failures during their reasoning cycle. Our analysis indicates that the existing archit...

Yes Yes

Yes Yes

The importance of integrating rules and ontologies for the Semantic Web has been well addressed by many researchers. Defeasible Logic is a simple but efficient nonmonotonic language which can handle both defeasibility and priority. In this paper we propose a novel approach to combining Defeasible Logic with Description Logics by introducing the Des...

In recent years, there has been a large amount of disparate work concerning the representation and reasoning with qualitative preferential information by means of approaches to nonmonotonic reasoning. Given the variety of underlying systems, assumptions, motivations, and intuitions, it is difficult to compare or relate one approach with another. He...

Neural network ensemble can significantly improve the generalization ability of neural network based systems. However, its comprehensibility is even worse than that of a single neural network because it comprises a collection of individual neural networks. ...

In recent years there has been large amount of disparate work concerning the representation and reasoning with preferential information in approaches to nonmonotonic reasoning. Given the variety of underlying systems, assumptions, motivations, and intuitions, it is difficult to compare or relate one approach with another. Here we present an overvie...

We provide a semantic framework for preference handling in answer set programming. To this end, we introduce preference preserving consequence operators. The resulting fixpoint characterizations provide us with a uniform semantic framework for characterizing preference handling in existing approaches. Although our approach is extensible to other se...

We provide a semantic framework for preference handling in answer set programming. To this end, we introduce preference preserving consequence operators. The resulting fixpoint characterizations provide us with a uniform semantic framework for characterizing preference handling in existing approaches. Although our approach is extensible to other se...

Much work has been done on extending the well-founded semantics to general disjunctive logic programs and various approaches have been proposed. However, these semantics are different from each other and no consensus is reached about which semantics is the most intended. In this paper we look at disjunctive well-founded reasoning from different ang...

Part of this work was done while this Guest Editor was with Tsinghua University, Beijing

No Yes

Skepticism is one of the most important semantic intuitions in artificial intelligence. The semantics formalizing skeptical reasoning in (disjunctive) logic programming is usually named well-founded semantics. However, the issue of defining and computing the well-founded semantics for disjunctive programs and databases has proved to be far more com...