Domain Knowledge Acquisition Research Articles

Strategy of Extracting Domain Knowledge for STRIPS World Based on Derived Predicates

领域知识的获取是智能规划研究中的重要内容之一.派生规则是一种基于逻辑推理的领域知识表示方法.在对动作模型和派生规则综合分析的基础上提出了基于派生谓词的STRIPS 领域知识提取策略,并给出了该提取策略的算法描述.在规划求解过程中,利用提取所得的领域规则可减少派生规则的逻辑推导,从而提高规划效率.对任意一个规划领域,利用提取所得的领域规则可以分析出领域谓词之间的互斥关系,这种互斥关系可以对一类矛盾的规划状态进行判定.最后,把所提出的领域规则提取策略嵌入到规划器StepByStep 之中进行了领域规则提取实验.实验结果验证了该提取算法的可行性和有效性,所提取出的领域规则能够直观地表达谓词间的因果关系,为派生谓词的真值判断和后续的规划求解提供了可靠的领域知识.;Domain knowledge acquisition is essential in the AI planning. Derived rule is a representation to the domain knowledge based on logical reasoning. On the basis of the action model and derived rules analysis, this paper proposes a strategy of extracting domain knowledge for STRIPS world based on derived predicates and the algorithm GetDomainRule for the strategy. The domain rules extracted by the algorithm are used to reduce the logical deduction of derived rules, enhancing the efficiency of the planning. For any domain, the mutex properties of any domain can be obtained between predicates from the domain rules, applied in judging the inconsistent planning state. Finally, the strategy proposed in this paper is embedded in the planner StepByStep and experiments are conducted to extract domain rules. Experimental results show the feasibility and validity of the algorithm GetDomainRule, and the domain rules extracted by the algorithm express the causal relations between predicates directly, providing reliable domain knowledge for determining the true values of derived predicates and the following planning.

Enhancing collaborative learning through dynamic forms of support: the impact of an adaptive domain‐specific support strategy

AbstractResearch on computer‐supported collaborative learning (CSCL) has strongly emphasized the value of providing student support of either fixed (e.g. collaboration scripts) or dynamic form (e.g. adaptive supportive interventions). Currently, however, there is not sufficient evidence corroborating the potential of adaptive support methods to improve domain‐knowledge acquisition in collaborative activities. Furthermore, available adaptive CSCL systems are difficult to integrate in everyday instruction because most of them are research prototypes. This study investigates the capacity of an adaptive support strategy [implemented in the open source Learning Activity Management System (LAMS) environment] to further improve learning outcomes in the context of a scripted collaborative activity. Thirty‐six students collaborated remotely in dyads on a task structured by a collaboration script. Control group dyads worked without any additional support. Treatment dyads were supported by a domain‐specific adaptive intervention in the form of reminding prompts. Treatment students outperformed those in the control group in domain knowledge acquisition. Overall, this study provides evidence that (1) adaptive forms of domain‐specific support can substantially improve learning in the context of a scripted collaborative activity; and (2) implementing simple forms of adaptive support is feasible and can be easily implemented by instructors in a freely available collaboration support system (such as LAMS).

The value of writing-to-learn when using question prompts to support web-based learning in ill-structured domains

This study investigates the effectiveness of two variants of a prompting strategy that guides students to focus on important issues when learning in an ill-structured domain. Students in three groups studied individually Software Project Management (SPM) cases for a week, using a web-based learning environment designed especially for this purpose. The first group (control) studied the cases without any prompting. The second group (“writing mode”) studied the same cases, while prompted to provide written answers to a set of knowledge integration prompts meant to engage students in deeper processing of the material. The third group (“thinking mode”) studied the cases, while prompted only to think of possible answers to the same question prompts. Results indicated that students in the writing condition group outperformed the others in both domain knowledge acquisition and knowledge transfer post-test items. Several students in the thinking condition group skipped the question prompts, while those that reported having reflected on the material were unable to achieve high performance comparable to the writing condition group. Overall, the study provides evidence that the implementation of prompting techniques in technology-enhanced learning environments may lead to improved outcomes, when combined with the requirement that students provide their answers in writing.

A framework and computer system for knowledge-level acquisition, representation, and reasoning with process knowledge

The development of knowledge-based systems is usually approached through the combined skills of software and knowledge engineers (SEs and KEs, respectively) and of subject matter experts (SMEs). One of the most critical steps in this task aims at transferring knowledge from SMEs’ expertise to formal, machine-readable representations, which allow systems to reason with such knowledge. However, this process is costly and error prone. Alleviating such knowledge acquisition bottleneck requires enabling SMEs with the means to produce the target knowledge representations, minimizing the intervention of KEs. This is especially difficult in the case of complex knowledge types like processes. The analysis of scientific domains like Biology, Chemistry, and Physics uncovers: (i) that process knowledge is the single most frequent type of knowledge occurring in such domains and (ii) specific solutions need to be devised in order to allow SMEs to represent it in a computational form. We present a framework and computer system for the acquisition and representation of process knowledge in scientific domains by SMEs. We propose methods and techniques to enable SMEs to acquire process knowledge from the domains, to formally represent it, and to reason about it. We have developed an abstract process metamodel and a library of problem solving methods (PSMs), which support these tasks, respectively providing the terminology for SME-tailored process diagrams and an abstract formalization of the strategies needed for reasoning about processes. We have implemented this approach as part of the DarkMatter system and formally evaluated it in the context of the intermediate evaluation of Project Halo, an initiative aiming at the creation of question answering systems by SMEs.

The influence of leads on cognitive load and learning in a hypertext environment

The purpose of this study was to determine the effect of leads (or hypertext node previews) on cognitive load and learning. Leads provided a brief summary of information in the linked node, which helped orient the reader to the linked information. Dependent variables included measures of cognitive load: self-report of mental effort, reading time, and event-related desynchronization percentage of alpha, beta and theta brain wave rhythms; and learning performance: a recall task, and tests of domain and structural knowledge. Results indicated that use of leads reduced brain wave activity that may reflect split attention and extraneous cognitive load, and improved domain and structural knowledge acquisition. Further, findings provide insights into differentiating the types of cognitive load apparent in hypertext-assisted learning environments. Use of EEG measures allowed examination of instantaneous cognitive load, which showed that leads may be influencing germane load—reducing mental burden associated with creating coherence between two linked node. The self-report of mental effort measure appears more closely associated with overall and intrinsic load.

Constraint acquisition for Entity-Relationship models

We establish search algorithms from the area of propositional logic as invaluable tools for the semantic knowledge acquisition in the conceptual database design phase. The acquisition of such domain knowledge is crucial for the quality of the target database. Integrity constraints are conditions that capture the semantics of the application domain under consideration. They restrict the databases to those that are considered meaningful to the application at hand. In practice, the decision of specifying a constraint is very important and extremely challenging. We show how techniques from propositional logic can be utilised to offer decision support for specifying Boolean and multivalued dependencies between properties of entities and relationships in conceptual databases. In particular, we use a search version of SAT-solvers to semi-automatically generate sample databases for this class of dependencies in Entity-Relationship models. The sample databases enable design participants to judge, justify, convey and test their understanding of the semantics of the future database. Indeed, the decision by the participants to specify a dependency explicitly is reduced to their decision whether there is some sample database that they can accept as a future database instance.

Spatial decision support system for Medfly control in citrus

A spatial decision support system (SDSS), designated MedCila was developed for controlling Medfly on citrus in Israel. The development involved four main phases: (1) acquisition of relevant expert and domain knowledge; (2) identification of the relevant criteria and modeling each criterion and the overall decision-making procedure; (3) integration of the MedCila into a GIS environment; and (4) initial evaluation of MedCila performance. The criteria found to be most relevant for control decision-making were the number of flies and the presence of a ‘blue eye’ in the nearest trap, the host-species susceptibility, the relative development of the Medfly based on accumulative day-degree model, the history of trapping, and the Medfly population in the nearby traps. Binary, linear, logarithmic and biological-based models were developed for the criteria identified. The overall decision-making procedure of the MedCila was based on the Stanford Certainty Theory integrated with a rule-based decision tree. Initial evaluation of the MedCila performance was done by retrospective comparison between the MedCila recommendations and the coordinator decisions. It was shown that the MedCila provides recommendations that are generally accepted by the coordinators; it reduces the number of unnecessary spray actions in the absence of a Medfly threat in space and in time; and it reduces the number of plots for which the coordinator needs to make a decision for.

Innate Ideas Revisited: For a Principle of Persistence in Infants' Physical Reasoning

The notion of innate ideas has long been the subject of intense debate in the fields of philosophy and cognitive science. Over the past few decades, methodological advances have made it possible for developmental researchers to begin to examine what innate ideas-what innate concepts and principles-might contribute to infants ' knowledge acquisition in various core domains. This article focuses on the domain of physical reasoning and on Spelke's (1988, 1994) proposal that principles of continuity and cohesion guide infants' interpretation of physical events. The article reviews recent evidence that these two principles are in fact corollaries of a single and more powerful principle of persistence, which states that objects persist, as they are, in time and space.

The effect of scaffolding students’ context-generating cognitive activity in technology-enhanced case-based learning

This study investigates the hypothesis that students’ learning and problem-solving performance in ill-structured domains can be improved, if elaborative question prompts are used to activate students’ context-generating cognitive processes, during case study. Two groups of students used a web-based learning environment to criss-cross and study case-based material in the software project management domain. The experimental group was additionally prompted to consistently answer a set of questions based on a model of context-generating processes, meant to engage students in deeper processing of information presented in cases. Students were also classified as having either “complex” or “simple” EB profile (based on their epistemological beliefs record), thereby establishing a 2 × 2 factorial design. Results indicated that scaffolding treatment had a significant main effect on students’ performance, with the experimental group performing better in both domain knowledge acquisition and knowledge transfer post-test items. There is also tentative indication that EB profile and scaffolding treatment interact, with complex-EB learners benefiting most from the scaffolded condition. Overall, the study provides evidence that it is possible to improve individual learning in a technology environment for case-based learning, by implementing appropriate questioning strategies that trigger students to activate their context-generating cognitive processes, while studying the contextually rich material of cases.

Children's acquisition of science terms: Simple exposure is insufficient

The ability of school children ( N = 233) to acquire new scientific vocabulary was examined. Children from two age groups ( M = 4.8 and M = 6.5) were introduced to previously unknown words in an educational video. Word knowledge was assessed through accuracy and latency for production and comprehension over a 9-month period. A draw and write task assessed acquisition of domain knowledge. Word learning was poorer than has previously been reported in the literature, and subject to influences of word type (domain specificity) and word class. The results indicate that the acquisition of scientific terms is a complex process moderated by lexical, semantic and pragmatic factors.

A Delphi-based approach to developing expert systems with the cooperation of multiple experts

Knowledge acquisition has been a critical bottleneck in building knowledge-based systems. In past decades, several methods and systems have been proposed to cope with this problem. Most of these methods and systems were proposed to deal with the acquisition of domain knowledge from single expert. However, as multiple experts may have different experiences and knowledge on the same application domain, it is necessary to elicit and integrate knowledge from multiple experts in building an effective expert system. Moreover, the recent literature has depicted that “time” is an important parameter that might significantly affect the accuracy of inference results of an expert system; therefore, while discussing the elicitation of domain expertise from multiple experts, it becomes an challenging and important issue to take the “time” factor into consideration. To cope with these problems, in this study, we propose a Delphi-based approach to eliciting knowledge from multiple experts. An application on the diagnosis of Severe Acute Respiratory Syndrome has depicted the superiority of the novel approach.

Semi-automatic learning of simple diagnostic scores utilizing complexity measures

Knowledge acquisition and maintenance in medical domains with a large application domain ontology is a difficult task. To reduce knowledge elicitation costs, semi-automatic learning methods can be used to support the domain specialists. They are usually not only interested in the accuracy of the learned knowledge: the understandability and interpretability of the learned models is of prime importance as well. Then, often simple models are more favorable than complex ones. We propose diagnostic scores as a promising approach for the representation of simple diagnostic knowledge, and present a method for inductive learning of diagnostic scores. It can be incrementally refined by including background knowledge. We present complexity measures for determining the complexity of the learned scores. We give an evaluation of the presented approach using a case base from the fielded system SonoConsult. We further discuss that the user can easily balance between accuracy and complexity of the learned knowledge applying the presented measures. We argue that semi-automatic learning methods can support the domain specialist efficiently when building (diagnostic) knowledge systems from scratch. The presented complexity measures allow for an intuitive assessment of the learned patterns.

Age, Ability, and the Role of Prior Knowledge on the Acquisition of New Domain Knowledge: Promising Results in a Real-World Learning Environment.

Prior knowledge, fluid intelligence (Gf), and crystallized intelligence (Gc) were investigated as predictors of learning new information about cardiovascular disease and xerography with a sample of 199 adults (19 to 68 years). The learning environment included a laboratory multimedia presentation (high-constraint-maximal effort), and a self-directed at-home study component (low-constraint-typical performance). Results indicated that prior knowledge and ability were important predictors of knowledge acquisition for learning. Gc was directly related to learning from the video for both domains. Because the trajectory of Gc stays relatively stable throughout the life span, these findings provide a more optimistic perspective on the relationship between aging and learning than that offered by theories that focus on the role of fluid abilities in learning.

Effect of repetitive transcranial magnetic stimulation on rate of memory acquisition

Repetitive transcranial magnetic stimulation (rTMS) may temporarily accelerate knowledge acquisition by neural networks, possibly by promoting rapid Hebbian learning. The authors tested this hypothesis in 20 normal subjects by comparing the impact of 25 minutes of high-frequency left dorsolateral prefrontal rTMS with that of sham rTMS on subsequent knowledge acquisition in several procedural and declarative memory domains. No significant group effects, positive or negative, were noted for any memory acquisition test, but prefrontal rTMS did reduce motor evoked potential threshold.

Maximum entropy model-based baseball highlight detection and classification

In this paper, we propose a novel system that is able to automatically detect and classify baseball highlights by seamlessly integrating image, audio, and speech clues using a unique framework based on maximum entropy model (MEM). What distinguishes our system is that we emphasize on the integration of multimedia features and the acquisition of domain knowledge through automatic machine learning processes. Our MEM-based framework provides a simple platform capable of integrating multimedia features as well as their contextual information in a uniform fashion. Unlike the Hidden Markov Model and the Bayes Network-based approaches, this framework does not need to explicitly segment and classify the input video into states during its data modeling process, and hence remarkably simplifies the training data creation and the highlight detection/classification tasks. Experimental evaluations demonstrate the superiority of our proposed baseball highlight detection system in its capability of detecting more major baseball highlights, and in its overall accuracy of detecting these major highlights.

Skillful Coaching: New Directions In Teaching Health Assessment

The purpose of this study was to examine differences in nursing student’s assessment skills before and after the implementation of a cognitive apprenticeship didactic approach including think-aloud and critical dialogue. In the nursing care of clients, a through and accurate holistic health assessment is essential. The content of the assessment task is taught in all nursing programs. Students must be competent not only in performing the examination but also in evaluating the information obtained and integrating it with the health history and lab findings. This extensive criteria makes assessment a challenging skill to master. Problems arise when students in nursing attempt to apply the theoretical assessment approaches to the clinical setting with real clients. Consequently, student’s performance outcomes are often unsatisfactory, as the teaching format usually focuses on the students’ acquisition of domain knowledge and psychomotor skills rather than on shaping the thinking processes involved. Students’ common problems with physical assessment shows that they struggle to manipulate the abstract positions involved.

Individual differences in learning entrepreneurship and their implications for web‐based instruction in e‐business and e‐commerce

AbstractThis paper reports on outcomes from a White Rose Centre for Enterprise funded project Managing Innovation in the Digital Economy. This project aims to incorporate learning for enterprise into undergraduate and postgraduate curricula in information management and information systems. The paper presents results from a survey distributed to postgraduate information management and information systems students following completion of a module in E‐Business and E‐Commerce delivered by the Department of Information Studies at the University of Sheffield. The findings suggest that differences in levels of prior knowledge of business studies and in cultural background can impact on students’ acquisition of domain knowledge and intellectual and information research skills during collaborative development of a business plan. Implications of the results for web‐based instruction are addressed, by identifying teaching and learning strategies that support differential treatment of learners in terms of content and process. It is concluded that empirical investigation of individual differences within a student body can inform more effective methods for information systems and information management graduates to learn about and for entrepreneurship.

The acquisition of price-relevant domain knowledge by a market

Through the application of policy capturing techniques in an experimental market, we provide an initial investigation into the process whereby individual domain knowledge is combined into market domain knowledge. We conduct a series of two-phase experiments where subjects are presented with actual financial information based on 20 publicly traded companies. In the first phase, subjects make individual decisions regarding the price of each company's stock. In the second phase, they participate in a computerized double auction asset market trading shares in the same 20 companies. We then model each individual's information usage and identify individual knowledge functions as well as model the market's information usage and identify the market knowledge function. We find that the aggregate market knowledge function is well characterized by a simple averaging of the knowledge functions of the participants.

Special Issue of the journal Artificial Intelligence on Reformulation

Since the early days of Artificial Intelligence, it has been recognized that problem reformulation is central to human reasoning and the ability of computer systems to reason effectively in complex domains. Reformulations have been used in a variety of problem-solving settings including automatic programming, constraint satisfaction, design, diagnosis, machine learning, planning, qualitative reasoning, scheduling and theorem proving. The primary use in such settings has been to overcome computational intractability by decreasing the combinatorial costs associated with searching large spaces. In addition, reformulation techniques are useful for knowledge acquisition and explanation generation in complex domains. The considerable interest in reformulation has led to a series of successful workshops over the last few years. AAAI workshops in 1990 and 1992 focused on selecting, constructing and using abstractions and approximations, while a series of workshops in 1989, 1990 and 1992 focused on problem reformulations. There was considerable intersection in the set of attendees and topics of the two separate workshop series, and this lead to holding merged workshops in 1994, 1995, 1998 and 2000 (entitled “Symposia on Abstraction, Reformulation, and Approximation”). We are preparing a special issue of the journal Artificial Intelligence to bring together papers describing and extending the current state of the art in problem reformulation. Topics will include but not be limited to the following: • techniques for automatically generating problem reformulations; • methods for selecting which of several reformulations are best for a given problem; • frameworks that unify and classify reformulation techniques; • theoretical and/or empirical studies of the costs and benefits of reformulation; • applications of reformulation: – search, constraint satisfaction, planning, theorem proving, logic programming; – simulation, design, diagnosis and control of physical systems; – automatic programming, analogical-reasoning, case-based reasoning, machine learning, knowledge-compilation and speedup learning; – knowledge representation, common sense reasoning, information systems, knowledge management systems;

A lite bioinformatics specialization

The emerging bioinformatics research community can benefit from an increase in computer science collaborators. One of the barriers to computer science collaboration is the lack of bioinformatics domain knowledge in the computer science community. This paper presents a strategy for facilitating this domain knowledge acquisition with the goal of accelerating the collaboration development process, particularly on campuses in which the computer science and bioformatics/medical communities are co-resident.