Incremental Knowledge Acquisition Research Articles

A NEW APPROACH FOR THE INCREMENTAL DEVELOPMENT OF RETRIEVAL FUNCTIONS FOR CBR

We present a new approach to the effective development of complex retrieval components for case-based reasoning systems (CBR). Our approach goes beyond the traditional CBR approach by allowing an incremental refinement of an existing retrieval knowledge base during routine use of the system. The refinement takes place through a direct expert-system interaction while the expert is accomplishing their given tasks. We lend ideas from ripple-down rules (RDR), a proven method for the very effective and efficient acquisition of classification knowledge during the routine use of a knowledge-based system (KBS). In our approach the expert is only required to provide explanations of why, for a given problem, a certain case should be retrieved. Incrementally a complex retrieval knowledge base as a composition of many simple retrieval functions is developed. This approach is effective with respect to both the development of highly tailored and complex retrieval knowledge bases for CBR as well as providing an intuitive and feasible approach for the expert. The approach has been implemented in our CBR system MIKAS (Menu construction using an Incremental Knowledge Acquisition System) that allows to automatically construct a menu that is strongly tailored to the individual requirements and food preferences of a client.

A knowledge-based approach to assign breast cancer treatments in oncology units

The aim of this work is to present a developed knowledge-based approach for breast cancer treatment. This Decision Support System (DSS) uses an incremental knowledge acquisition technique called Multiple Classification Ripple Down Rules (MCRDR) and a breast cancer treatments knowledge base. This system integrate MCRDR inference engine and use knowledge bases to reach advanced reasoning level during the decision making process. The knowledge base will be built from clinical protocols, and its content will be updated by the doctor using the MCRDR component. This DSS infers a cancer treatment from the (clinical or pathological) input data supplied by the physician and it also allows the maintenance of the knowledge base.

Addressing the Ontology Acquisition Bottleneck Through Reverse Ontological Engineering

The use of ontologies in knowledge engineering arose as a solution to the difficulties associated with acquiring knowledge, commonly referred to as the knowledge acquisition bottleneck. The knowledge-level model represented in an ontology provides a much more structured and principled approach compared with earlier transfer-of-symbolic-knowledge approaches but brings with it a new problem, which can be termed the ontology-acquisition (and maintenance) bottleneck. Each ontological approach offers a different structure, different terms and different meanings for those terms. The unifying theme across approaches is the considerable effort associated with developing, validating and connecting ontologies. We propose an approach to engineering ontologies by retrospectively and automatically discovering them from existing data and knowledge sources in the organization. The method offered assists in the identification of similar and different terms and includes strategies for developing a shared ontology. The approach uses a human-centered, concept-based knowledge processing technique, known as formal concept analysis, to generate an ontology from examples. To assist classification of examples and to identify the salient features of the example, we use a rapid and incremental knowledge acquisition and representation technique, known as ripple-down rules. The method can be used as an alternative or complement to other approaches.

Adaptive Ripple Down Rules Method based on Description Length

A knowledge acquisition method Ripple Down Rules (RDR) can directly acquire and encode knowledge from human experts. It is an incremental acquisition method and each new piece of knowledge is added as an exception to the existing knowledge base. Past researches on RDR method assume that the problem domain is stable. This is not the case in reality, especially when an environment changes. Things change over time. This paper proposes an adaptive Ripple Down Rules method based on the Minimum Description Length Principle aiming at knowledge acquisition in a dynamically changing environment. We consider the change in the correspondence between attribute-values and class labels as a typical change in the environment. When such a change occurs, some pieces of knowledge previously acquired become worthless, and the existence of such knowledge may hinder acquisition of new knowledge. In our approach knowledge deletion is carried out as well as knowledge acquisition so that useless knowledge is properly discarded to ensure efficient knowledge acquisition while maintaining the prediction accuracy for future data. Furthermore, pruning is incorporated into the incremental knowledge acquisition in RDR to improve the prediction accuracy of the constructed knowledge base. Experiments were conducted by simulating the change in the correspondence between attribute-values and class labels using the datasets in UCI repository. The results are encouraging.

An approach for incremental knowledge acquisition from text

Abstract The approach presented in this work aims to simplify Knowledge Acquisition Processes by means of the extraction of some types of knowledge directly from natural language texts. This approach uses both a morphologic analyzer and a user-centered, incremental knowledge methodology with the purpose of achieving a language independent system. The knowledge acquired from text is represented by means of ontologies.

Building a case-based diet recommendation system without a knowledge engineer

We present a new approach to the effective development of menu construction systems that allow to automatically construct a menu that is strongly tailored to the individual requirements and food preferences of a client. In hospitals and other health care institutions dietitians develop diets for clients which need to change their eating habits. Many clients have special needs in regards to their medical conditions, cultural backgrounds, or special levels of nutrient requirements for better recovery from diseases or surgery, etc. Existing computer support for this task is insufficient—many diets are not specifically tailored for the client’s needs or require substantial time of a dietitian to be manually developed. Our approach is based on case-based reasoning, an artificial intelligence technique that finds increasing entry into industrial practice. Our approach goes beyond the traditional case-based reasoning (CBR) approach by allowing an incremental improvement of the system’s competency during routine use of the system. The improvement of the system takes place through a direct expert user–system interaction while the expert is accomplishing their tasks of constructing a diet for a given client. Whenever the system performs unsatisfactorily, the expert will need to modify the system-produced diet ‘manually’, i.e. by entering the desired modifications into the system. Our implemented system, menu construction using an incremental knowledge acquisition system (MIKAS), asks the expert for simple explanations for each of the manual actions he/she takes and incorporates the explanations automatically into its knowledge base (KB) so that the system will perform these manually conducted actions automatically at the next occasion. We present MIKAS and discuss the results of our case study. While still being a prototype, the senior clinical dietitian involved in our evaluation studies judges the approach to have considerable potential to improve the daily routine of hospital dietitians as well as to improve the average quality of the dietary advice given to patients within the limited available time for dietary consultations. Our approach opens up a new avenue towards building highly specialised CBR systems in a more cost-effective way. Hence, our approach promises to allow a significantly more widespread development and practical deployment of CBR systems in a large variety of application domains including many medical applications.

An advanced artificial intelligence tool for menu design.

The computer-assisted menu design still remains a difficult task. Usually knowledge that aids in menu design by a computer is hard-coded and because of that a computerised menu planner cannot handle the menu design problem for an unanticipated client. To address this problem we developed a menu design tool, MIKAS (menu construction using incremental knowledge acquisition system), an artificial intelligence system that allows the incremental development of a knowledge-base for menu design. We allow an incremental knowledge acquisition process in which the expert is only required to provide hints to the system in the context of actual problem instances during menu design using menus stored in a so-called Case Base. Our system incorporates Case-Based Reasoning (CBR), an Artificial Intelligence (AI) technique developed to mimic human problem solving behaviour. Ripple Down Rules (RDR) are a proven technique for the acquisition of classification knowledge from expert directly while they are using the system, which complement CBR in a very fruitful way. This combination allows the incremental improvement of the menu design system while it is already in routine use. We believe MIKAS allows better dietary practice by leveraging a dietitian's skills and expertise. As such MIKAS has the potential to be helpful for any institution where dietary advice is practised.

Theoretical basis for hierarchical incremental knowledge acquisition

Human experts tend to introduce intermediate terms in giving their explanations. The expert's explanation of such terms is operational for the context that triggered the explanation; however, term definitions remain often incomplete. Further, the expert's (re) use of these terms is hierarchical (similar to natural language). In this paper, we argue that a hierarchical incremental knowledge acquisition (KA) process that captures the expert terms and operationalizes them while incompletely defined makes the KA task more effective. Towards this we present our knowledge representation formalism Nested Ripple Down Rules (NRDR) that is a substantial extension to the (Multiple Classification) Ripple Down Rule (RDR) KA framework. The incremental KA process with NRDR as the underlying knowledge representation has confirmation holistic features. This allows simultaneous incremental modelling and KA and eases the knowledge base (KB) development process.Our NRDR formalism preserves the strength of incremental refinement methods, that is the ease of maintenance of the KB. It also addresses some of their shortcomings: repetition, lack of explicit modelling and readability. KBs developed with NRDR describe an explicit model of the domain. This greatly enhances the reuseability of the acquired knowledge.This paper also presents a theoretical framework for analysing the structure of RDR in general and NRDR in particular. Using this framework, we analyse the conditions under which RDR converges towards the target KB. We discuss the maintenance problems of NRDR as a function of this convergence. Further, we analyse the conditions under which NRDR offers an effective approach for domain modelling. We show that the maintenance of NRDR requires similar effort to maintaining RDR for most of the KB development cycle. We show that when an NRDR KB shows an increase in maintenance requirement in comparison with RDR during its development, this added requirement can be automatically handled using stored past seen cases.

Logistic Chaos Protects Evolution against Environment Noise

We propose a neuron with noise generator specially for the evolutionary robotics approach to incremental knowledge acquisition. Genetically evolving neural networks are modified continuously by genetic operations. Difficulty in incrementing knowledge when a neural network acts as a robotic controller arises when network operates unlike in the past due to disturbance by neurons added by genetic operators. To evolve a network robust against such internal noise, we propose adding noise generators to neurons. We show the effectiveness of the application of a logistic chaos noise generator to neurons by comparing several noise generators.

Incremental knowledge acquisition in supervised learning networks

Acquiring new knowledge without interfering with old knowledge is a key issue in designing an incremental-learning system. The success of such a system hinges on an embedded incrementable information structure with improved performance over time. This paper describes an incremental-learning network for pattern recognition that uses a rule-based connectionist technique to represent general domain and case-specific knowledge, uses bounded weight modification to update its connection weights, and also performs structural learning. Specific strategies are developed for preventing overtraining and for incrementally growing and pruning the network. The soundness of this approach is demonstrated by empirical studies in two independent domains.

Semi-autonomous evolution of object models for adaptive object recognition

The paper presents a semi-autonomous model evolution approach to object recognition under variable perceptual conditions. The approach assumes that (i) the system has to recognize objects on separate images of a sequence, and (ii) the images demonstrate the variability of conditions under which objects are perceived (gradual change in resolution, lighting, positioning). The adaptation of object models is executed due to perceived, over a sequence of images, variabilities of object characteristics. This adaptation involves (i) the application of learned models to the next image, (ii) the monitoring of recognition effectiveness of the models, and (iii) an activation of learning processes if needed (i.e., when the recognition effectiveness of the models decreases). Model adaptation (evolution) integrates recognition processes of computer vision with incremental knowledge acquisition processes of machine learning in a closed loop. The paper presents both an outline of the iterative evolution methodology and the investigation of an incremental model generalization approach using the example of a texture recognition problem. Experiments were run in a semi-autonomous mode where a teacher secured soundness behavior of the evolution system. The experiments are compared for three system configurations: (i) a one-level control structure, (ii) a two-level control structure, and (iii) a two-level control structure with data filtering. The obtained results are evaluated for system recognition effectiveness, recognition stability, and predictability of evolved models.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

LBS: Bayesian learning system for rapid expert system development

Abstract This paper presents the results of experience with a novel expert system shell called Learning Base System (LBS). The primary goal in developing LBS was to enable the rapid development of testable expert systems. The strategy adopted was to use a Bayesian classifier system as the form of knowledge representation, and adapt it to allow incremental acquisition-of knowledge from both data and experts, and prediction and validation procedures. The advantages and limitations of the system are described in three applications. The first application is the diagnosis of diseases in crops, illustrating knowledge acquisition by an expert in a data-poor domain. The second illustrates how LBS could be used in a geographic information system. The third is the development and testing of models for predicting wildlife density solely from data. The Bayesian classifier is shown to be a flexible formalism for implementing a wide variety of knowledge-based tasks.

A knowledge acquisition inductive system driven by empirical interpretation of derived results

This paper describes the design philosophy of and current issues concerning a knowledge acquisition system namedkaiser. This system is an intelligent workbench for construction of knowledge bases for classification tasks by domain experts themselves. It first learns classification knowledge inductively from the examples given by a human expert, then analyzes the result and process based on abstract domain knowledge which is also given by the expert. Based on this analysis, it asks sophisticated questions for acquiring new knowledge. The queries stimulate the human expert and help him to revise the learned results, control the learning process and prepare new examples and domain knowledge. Viewed from an AI aspect, it aims at integrating similarity-based inductive learning and explanation-based deductive reasoning by guiding inductive inference with theoretical and/or heuristic knowledge about the domain. This interactive induce-evaluate-ask cycle produces a rational interview which promotes incremental acquisition of domain knowledge as well as efficient induction of operational and reasonable knowledge proved by the domain knowledge.