Algorithms for Merging Probabilistic Knowledge Bases

Resolving the inconsistency that appears in knowledge bases is an extremely important stage in the process of merging knowledge bases. With probabilistic context, many studies have solved the inconsistency of a probabilistic knowledge base (PKB) with different approaches. However, these previous focus on addressing the problem of restoring consistency in a PKB by employing the principles of maximum entropy or building the consistency restorer with complex creeping functions. We discover that it is really difficult to find the functions that are utilized in the creeping consistency restorers. Therefore, the process of finding these functions may lead to a poor performance. Based on that, we proposed the inconsistency solvers for a PKB, where we consider revising the functions involved in handling the inconsistency in a simpler direction. Two the inconsistency solvers are the equitable deformation inconsistency solver and the amerced deformation inconsistency solver. Corresponding to each solver, an algorithm to solve inconsistencies in a PKB is proposed. Moreover, the cost of each algorithm also are considered and proved.

We present the ARCHIMEDES system for efficient query processing over probabilistic knowledge bases. We design ARCHIMEDES for knowledge bases containing incomplete and uncertain information due to limitations of information sources and human knowledge. Answering queries over these knowledge bases requires efficient probabilistic inference. In this paper, we describe ARCHIMEDES's efficient knowledge expansion and querydriven inference over UDA-GIST, an in-database unified data- and graph-parallel computation framework. With an efficient inference engine, ARCHIMEDES produces reasonable results for queries over large uncertain knowledge bases. We use the Reverb-Sherlock andWikilinks knowledge bases to show ARCHIMEDES achieves satisfactory quality with real-time performance.

Evaluating the success of a knowledge acquisition (KA) task is difficult and expensive. Most evaluation approaches rely on the expert themselves, either directly, or indirectly by relying on data previously prepared with the help of experts. In incremental KA, knowledge base (KB) errors are monitored and corrected by an expert. Thus, during its evolution a record of the knowledge based system (KBS) performance is usually easy to keep. We propose to integrate with the incremental KA process, an evaluation process based on a statistical analysis to estimate the effectiveness of the KBS, as the KBS is actually evolved. We tailor such an analysis for Ripple Down Rules (RDR), which is an effective incremental KA methodology where a record of the KBS performance can be easily derived and updated as new cases are processed by the system. An RDR KB is a collection of rules with hierarchical exceptions, which are entered and validated by the expert in the context of their use. This greatly facilitates the knowledge maintenance task which, characteristically in RDR, overlaps with the incremental KA process. The work in this paper aims to overlap evaluation with maintenance and development of the knowledge base. It also minimises the major expense in deploying the RDR KBS, that of keeping a domain expert on-line during maintenance and the initial period of deployment. The expert is not kept on-line longer than it is absolutely necessary. We use the structure and semantics of an evolving RDR KB, combined with proven machine learning statistical methods, to estimate the added value in every KB update, as the KB evolves. Using these values, the decisionmakers in the organisation employing the KBS can apply a cost-benefit analysis of the continuation of the incremental KA process. They can then determine when this process, involving keeping an expert on-line, should be terminated.

Information extraction and human collaboration techniques are widely applied in the construction of web-scale knowledge bases. However, these knowledge bases are often incomplete or uncertain. In this paper, we present ProbKB, a probabilistic knowledge base designed to infer missing facts in a scalable, probabilistic, and principled manner using a relational DBMS. The novel contributions we make to achieve scalability and high quality are: 1) We present a formal definition and a novel relational model for probabilistic knowledge bases. This model allows an efficient SQL-based inference algorithm for knowledge expansion that applies inference rules in batches; 2) We implement ProbKB on massive parallel processing databases to achieve further scalability; and 3) We combine several quality control methods that identify erroneous rules, facts, and ambiguous entities to improve the precision of inferred facts. Our experiments show that ProbKB system outperforms the state-of-the-art inference engine in terms of both performance and quality.

Research in verification and validation (V&V) of knowledge based systems (KBS) has emerged as a distinct field in the last decade and is intended to address issues associated with KBS quality aspects and to credit such applications to the same degree as conventional applications. Part of the knowledge and techniques developed are now ready to be transferred to industry for the implementation of software certification. Certification is probably the most advanced use of the V&V methods, techniques and knowledge that have been developed so far. The paper approaches the challenging issues of certifying the knowledge base (KB) component of a KBS. KB certification uses a number of different certification methods such as static analysis, testing, inspection, and modeling to assess the quality of the KB component. Although these methods are currently used in KBS development, it is not yet clear how they would be applied by an independent certification organization to arrive at a pass/fail decision. The article discusses the components of KB certification and identifies application areas directly connected with industrial needs.

It is very difficult to build a knowledge base in the mechanical engineering field. Because there is a great deal of complicated knowledge in mechanical engineering in comparison with other engineering fields. To build a large knowledge base, much research to connect Data Base System (DBS) and Knowledge Base System (KBS) systematically has been conducted. DBS is suitable for storing much knowledge but is not suitable for the representation of knowledge. Therefore, we believe it is possible in the mechanical engineering field to realize DBS characters in KBS by using an extended deamon function and extended system slot. If this system is realized, KBS will be suitable for managing and representing knowledge, and it will be easy to build a knowledge base in the mechanical engineering field. This paper describes the method of building a large knowledge base in mechanical engineering using an extended daemon function and an extended system slot, and "the realization of such a knowledge base."

The article uses an example of a diagnostic system based on a probabilistic knowledge base and Naive Bayes Classifier inference, analyzing mismatch characteristic of the system’s reasoning and an expert’s reasoning to show the solution extending basic knowledge base with specialized, domain-dependent meta-knowledge. The evaluation of the solution is represented by the expert system dedicated to recognition of hypertension.

Artificial intelligence can assist in industrial control in many ways, but this paper concentrates on the role of knowledge-based systems in the areas of control, detection and diagnosis of faults, scheduling and planning. The paper argues that the application of knowledge-based techniques can provide a unification of these and other activities, and outlines the requirements of knowledge representation and inference techniques that are needed to make such systems a reality. The advances in hardware, including sensors as well as computers, that will be driven by and in turn stimulate the introduction of the knowledge-based industrial control systems are also considered.

Ever since MYCIN introduced the idea of computer-based explanations to the artificial intelligence community, it has come to be taken for granted that all knowledge-based systems (KBS) need to provide explanations. The first part of this dissertation proposes a cognitive learning theory based model that both clarifies the reasons as to why KBS need to provide explanations and serves as the basis for conceptualizing the provision of KBS explanations. Using the concepts of the feedforward and feedback operators of cognitive learning it develops strategies for providing KBS explanations and uses them to classify the various types of explanations found in current KBS applications. The roles of feedforward and feedback explanations within the context of the theory of cognitive skill acquisition and a model of expert judgment are also analyzed. These, together with past studies of KBS explanations, suggest that user expertise, the types of explanations provided, and the level of user agreement are significant factors that influence the explanation seeking behavior of users. The dissertation also explores the effects of the use of KBS explanations in judgmental decision making situations supported by a KBS. It identifies and considers four distinct categories of potential effects of the use of explanations--learning effects, perceived effects, behavioral effects, and effects on judgmental decision making. The second part of the dissertation empirically evaluates the explanation provision strategies in a laboratory experiment in which 80 novice and expert subjects used a KBS for financial analysis to make judgments under conditions of uncertainty. The experiment was designed specifically to investigate the following fundamental research questions: (1) To what extent are the various kinds of explanations used? (2) How does user expertise, the feedforward and feedback provision of explanations, and the level of user agreement influence the amount and the types of explanations that are used? and (3) Does the use of explanations affect the accuracy of judgmental decision-making and user perceptions of usefulness? Some of the major results relating to the determinants of the use of KBS explanations include: (1) user expertise is not a determinant of the proportion of explanations used but influences the types of explanations that are used, (2) explanation provision strategy is a critical determinant of the use of KBS explanations with feedback explanations being used significantly more than feedforward explanations, and (3) the three types of explanations are used in different proportions with the Why and How explanations being used significantly more that the Strategic explanations. (Abstract shortened by UMI.)

Knowledge-based systems (KBS) are being widely used for risk analysis of structures. The KBS make use of various types of observations made during postdisaster damage surveys and also the expertise of various domain engineers. Consolidation of all the observations to form the knowledge base (KB) is a difficult task. Each of the observations consists of various parameters. Establishing parameter interdependencies is important for creating a KB. The use of a graph theoretic technique is proposed in this paper to determine the interdependencies among various parameters useful in the creation of a KB. This approach helps in identifying necessary and sufficient data required to develop an effective KBS for risk analysis of structures. Using the proposed approach, an example problem of establishing parameter interdependencies for risk analysis of roof structure against damage due to cyclonic winds is solved. This example also demonstrates how sufficiency and consistency of the observations can be checked.

Knowledge-based Systems and Expert Systems, in particular, are expensive to build and difficult to validate and debug because of their complexity and dynamism. Therefore, it is not easy for knowledge engineer and domain expert to identify the gaps and mistakes in knowledge base. Unit testing is unable to cover validation process at all stages, in many cases manual thorough review of decision process is needed. In this paper we spot main approaches to validation and verification issue and describe a component that helps to debug a knowledge base by visualising execution of rules that derive a particular result. This component is developed for Knowledge-based Systems built on Drools Platform1 and we demonstrate application of this component in a knowledge-based engineering system for structural optical design.

The development of a knowledge-based software package for the analysis and control of electric distribution networks is discussed. A relational knowledge base, implemented in Prolog, is used to insure fast access to the data and an efficient post-reconfiguration update algorithm. An expert system shell DECIDE, is used to interpret human knowledge coded as knowledge clusters, and to provide a menu-driven user interface. The knowledge-based system described here has the following advantages: its expert system shell provides 'backup' and 'explanation' facilities; it can generate all the possible network reconfigurations and recommend the best choice; and the Prolog implementation of the algorithms is easy to maintain. The knowledge base is easy to expand and can be modified during execution. The prototype system demonstrates that artificial intelligence techniques can be used in power system analysis and reconfiguration, and provides a platform on which a base future research on the utility of expert system technology in distribution automation. >

Knowledge-based systems (KBS) are not necessarily based on a well-defined ontologies. In particular it is possible to build very successful KBS for classification problems, but where the classes or conclusions are entered by experts as free-text sentences with little constraint on textual consistency and little systematic organisation of the conclusions. This paper investigates how relations between such ‘classes’ may be discovered from existing knowledge bases. We have based our approach on KBS built with Ripple Down Rules (RDR). RDR is a knowledge acquisition and knowledge maintenance method which allows KBS to be built very rapidly and simply by correcting errors, but does not require a strong ontology. Our experimental results are based on a large real-world medical RDR KBS. The motivation for our work is to allow an ontology in a KBS to ‘emerge’ during development, rather than requiring the ontology to be established prior to the development of the KBS. It follows earlier work on using Formal Concept Analysis (FCA) to discover ontologies in RDR KBS.KeywordsKnowledge AcquisitionConcept LatticeClass RelationFormal ContextFormal Concept AnalysisThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

The implementation of security results in a user being restricted to a subset of a knowledge base. The information content of constraints may mean that a user is not permitted to access some constraints. But without access permission to all constraints, a user may make modifications which do not satisfy the integrity of other users in the knowledge base system. The author describes an approach which shows how users who have access to a subset of the knowledge base which does not satisfy all constraints can act to restore their subset to a state where the information satisfies all integrity constraints. The user may make modifications to restore integrity, may place information in a pending state, or may work with unaffected areas of the knowledge base. >

Dutch research on knowledge-based instructional systems: Introduction to the special issue