Software Interface Evaluation: Modeling of Human Error

1993 ◽  
Vol 37 (5) ◽  
pp. 453-455 ◽  
Author(s):  
S. J. Wright ◽  
S. J. Packebush ◽  
D. A. Mitta
Keyword(s):  
Human Error ◽  
Interface Element ◽  
Trial And Error ◽  
Rule Based ◽  
High Frequencies ◽  
Software Interface ◽  
Knowledge Based ◽  
Cognitive Levels ◽  
Graphics Software ◽  
And Task

The purpose of this study was to use a human error model to evaluate a commercially available Macintosh-based graphics application based upon the frequencies and types of mistakes occurring during users' performance of designated tasks. The occurrence of high frequencies of knowledge-based and rule-based mistakes during the learning of an interface element would indicate that the element requires evaluation and possible redesign. This study involved five participants, all of whom were students at Texas A&M University. The participants were experienced Macintosh users with no experience using Macintosh graphics software. The graphics environment of interest was MacDraw II® 1.0 Version 2 (Schutten, Goldsmith, Kaptanoglu, and Spiegel, 1988). Ten drawings created with the program were used to examine participants' cognitive levels and types of errors made throughout the process of familiarizing themselves with this program. The first drawing was created to exemplify simple figures created with the graphics tools in the program to illustrate shading. The second through tenth drawings incorporated these figures in several arrangements. All drawings incorporated eight tools (or tasks), and each tool was used only once in each drawing. The results indicated significant differences in frequencies of error types, frequencies of errors between tasks and frequencies of errors between trials. There were also interactions between trial and error, and task and error.

A Comparison of Rule-Based and Machine Learning Models for Classification of Human Factors Aviation Safety Event Reports

2020 ◽  
Vol 64 (1) ◽  
pp. 129-133
Author(s):  
Katherine Darveau ◽  
Daniel Hannon ◽  
Chad Foster
Keyword(s):  
Machine Learning ◽  
Human Factors ◽  
Human Error ◽  
Data Science ◽  
Aircraft Engine ◽  
Rule Based ◽  
Root Cause ◽  
Textual Data ◽  
Safety Event

There is growing interest in the study and practice of applying data science (DS) and machine learning (ML) to automate decision making in safety-critical industries. As an alternative or augmentation to human review, there are opportunities to explore these methods for classifying aviation operational events by root cause. This study seeks to apply a thoughtful approach to design, compare, and combine rule-based and ML techniques to classify events caused by human error in aircraft/engine assembly, maintenance or operation. Event reports contain a combination of continuous parameters, unstructured text entries, and categorical selections. A Human Factors approach to classifier development prioritizes the evaluation of distinct data features and entry methods to improve modeling. Findings, including the performance of tested models, led to recommendations for the design of textual data collection systems and classification approaches.

Ultrasonic Tomography for Rail Flaw Imaging

2014 ◽  
Author(s):  
Robert Phillips ◽  
Francesco Lanza di Scalea ◽  
Claudio Nucera ◽  
Piervincenzo Rizzo ◽  
Leith Al-Nazer
Keyword(s):  
Human Error ◽  
Ultrasonic Inspection ◽  
Quantitative Information ◽  
Inspection Time ◽  
Practical Implementation ◽  
Ultrasonic Tomography ◽  
Railroad Industry ◽  
Knowledge Based ◽  
Portable System ◽  
High Resolution Images

There is a need in the railroad industry to have quantitative information on internal rail flaws, including flaw size and orientation. Such information can lead to knowledge-based decision making on any remedial action, and ultimately increase the safety of train operations by preventing derailments. Current ultrasonic inspection methods leave such sizing determinations to the inspector, and there can be significant variability from one inspector to another depending on experience and other factors. However, this quantitative information can be obtained accurately by 3-D imaging of the rail flaws. It is the goal of this project to develop a portable system that will improve defect classification in rails and ultimately improve public safety. This paper will present a method for 3-D imaging of internal rail flaws based on Ultrasonic Tomography. The proposed technique combines elements of ultrasonic testing with those of radar and sonar imaging to obtain high-resolution images of the flaws using a stationary array of ultrasonic transducers. The array is operated in a “full matrix capture” scheme that minimizes the number of ultrasonic transmitters, hence simplifying the practical implementation and reducing the inspection time. In this method, a full 3D image of the rail volume identifies the location, size and orientation of the defect. This will help to eliminate human error involved with a typical manual inspection using a single transducer probe inspection. The results of advanced numerical simulations, carried out on a rail profile, will be presented. The simulations show the effectiveness of the technique to image a 5% Head Area Transverse Defect in the railhead. Current efforts are aimed at developing an experimental prototype based on this technology, whose design status is also discussed in this paper.

scholarly journals Comparing continual task learning in minds and machines

2018 ◽  
Vol 115 (44) ◽  
pp. E10313-E10322 ◽  
Author(s):  
Timo Flesch ◽  
Jan Balaguer ◽  
Ronald Dekker ◽  
Hamed Nili ◽  
Christopher Summerfield
Keyword(s):  
Human Error ◽  
Human Performance ◽  
Deep Neural Networks ◽  
Artificial Agents ◽  
Trial And Error ◽  
Stimulus Space ◽  
Cognitive Mechanisms ◽  
Error Training ◽  
Multiple Tasks ◽  
Continual Learning

Humans can learn to perform multiple tasks in succession over the lifespan (“continual” learning), whereas current machine learning systems fail. Here, we investigated the cognitive mechanisms that permit successful continual learning in humans and harnessed our behavioral findings for neural network design. Humans categorized naturalistic images of trees according to one of two orthogonal task rules that were learned by trial and error. Training regimes that focused on individual rules for prolonged periods (blocked training) improved human performance on a later test involving randomly interleaved rules, compared with control regimes that trained in an interleaved fashion. Analysis of human error patterns suggested that blocked training encouraged humans to form “factorized” representation that optimally segregated the tasks, especially for those individuals with a strong prior bias to represent the stimulus space in a well-structured way. By contrast, standard supervised deep neural networks trained on the same tasks suffered catastrophic forgetting under blocked training, due to representational interference in the deeper layers. However, augmenting deep networks with an unsupervised generative model that allowed it to first learn a good embedding of the stimulus space (similar to that observed in humans) reduced catastrophic forgetting under blocked training. Building artificial agents that first learn a model of the world may be one promising route to solving continual task performance in artificial intelligence research.

INTERACTION BETWEEN AGENTS IN ROUTINE, FAMILIAR AND UNFAMILIAR SITUATION

1996 ◽  
Vol 05 (01) ◽  
pp. 1-25 ◽  
Author(s):  
B. CHAIB-DRAA
Keyword(s):  
Multiagent System ◽  
Urban Traffic ◽  
Rule Based ◽  
Complex Tasks ◽  
Knowledge Based ◽  
Complex Interactions ◽  
Low Levels

A framework for designing a Multiagent System (MAS) in which agents are capable of coordinating their activities in routine, familiar, and unfamiliar situations is proposed. This framework is based on the Skills, Rules and Knowledge (S-R-K) taxonomy of Rasmussen. Thus, the proposed framework should allow agents to prefer the lower skill-based and rule-based levels rather than the higher knowledge-based level because it is generally easier to obtain and maintain coordination between agents in routine and familiar situations than in unfamiliar situations. The framework should also support each of the three levels because complex tasks combined with complex interactions require all levels. To permit agents to rely on low levels, a suggestion is developed: agents are provided with social laws so as to guarantee coordination between agents and minimize the need for calling a central coordinator or for engaging in negotiation which requires intense communication. Finally, implementation and experiments demonstrated, on some scenarios of urban traffic, the applicability of major concepts developed in this article.

scholarly journals Human Failure: An Analysis of 2000 Incident Reports

1993 ◽  
Vol 21 (5) ◽  
pp. 678-683 ◽  
Author(s):  
J. A. Williamson ◽  
R. K. Webb ◽  
A. Sellen ◽  
W. B. Runciman ◽  
J. H. Van Der Walt
Keyword(s):  
Crisis Management ◽  
Human Error ◽  
Best Practice ◽  
Contributing Factors ◽  
Rule Based ◽  
Relative Contribution ◽  
Technical Errors ◽  
Human Failure ◽  
Traditional Patterns ◽  
The Individual

Information of relevance to human failure was extracted from the first 2,000 incidents reported to the Australian Incident Monitoring Study (AIMS). All reports were searched for human factors amongst the “factors contributing”, “factors minimising”, and “suggested corrective strategies” categories, and these were classified according to the type of human error with which they were associated. In 83% of the reports elements of human error were scored by reporters. “Knowledge-based errors” contributed directly to about one-quarter of incidents; the outcome of one third of incidents was thought to have been minimised by prior experience or awareness of the potential problems, and in one fifth some strategy to improve knowledge was suggested. Correction of “rule-based errors” or provision of protocols or algorithms were thought, together, to have a potential impact on nearly half of all incidents. Failure to check equipment or the patient contributed to nearly one-quarter of all incidents, and inadequate crisis management contributed to a further I in 8. “Skill-based errors” (slips and lapses) were directly responsible for I in 10 of all incidents, and were thought to make an indirect contribution in up to one quarter. “Technical errors” were responsible for about 1 in 8 incidents. Analysing the relative contribution of each type of error for each type of problem allows the development of rational preventative strategies. Continued efforts must be made to improve the knowledge-base of anaesthetists, but AIMS has shown that there may also be much to gain from directing attention towards eliminating rule-based errors, for promoting the use of protocols, check-lists and crisis management algorithms, and improving anaesthetists’ insight into the factors contributing and circumstances in which slips and lapses may occur. Traditional patterns of behaviour in doctors may also make them more liable to certain types of human error; removing the onus for adhering to standards and approved work practices from the individual to the “system” may lead to more consistent application of the “best practice”.

Parallelism in knowledge-based machines

1989 ◽  
Vol 4 (1) ◽  
pp. 53-71
Author(s):  
Apostolos N. Refenes
Keyword(s):  
Expert Systems ◽  
Programming Languages ◽  
Parallel Machine ◽  
Computer Architectures ◽  
Parallel Computer ◽  
Rule Based ◽  
Boltzmann Machines ◽  
Knowledge Based ◽  
Parallel Tasks ◽  
Language Classes

AbstractThe application area of knowledge-based expert systems is currently providing the main stimulus for developing powerful, parallel computer architectures. Languages for programming knowledge-based applications divide into four broad classes: Functional languages (e.g. LISP), Logic languages (e.g. PROLOG), Rule-Based languages (e.g. OPS5), and, what we refer to as self-organizing networks (e.g. BOLTZMANN machines).Despite their many differences, a common problem for all language classes and their supporting machine architectures is parallelism: how to de-compose a single computation into a number of parallel tasks that can be distributed across an ensemble of processors. The aim of this paper is to review the four types of language for programming knowledge-based expert systems, and their supporting parallel machine architectures. In doing so we analyze the concepts and relationships that exist between the programming languages and their parallel machine architectures in terms of their strengths and limitations for exploiting parallelization.

An analysis of aircraft maintenance incidents using psychological and cognitive engineering knowledge

2016 ◽  
Vol 60 (1) ◽  
pp. 1676-1680
Author(s):  
Man-wa Ng ◽  
Simon Y. W. Li
Keyword(s):  
Data Entry ◽  
Cognitive Engineering ◽  
Aircraft Maintenance ◽  
Main Task ◽  
Rule Based ◽  
Knowledge Based ◽  
Data Entry Error ◽  
Starting Point ◽  
Analysis System ◽  
Error Types

The aim of the current analysis is to complement existing studies of aircraft maintenance incidents by providing finer and more detailed explanations for their causes in terms of task and error types. A total of 109 aircraft maintenance incidents were analyzed with respect to knowledge and concepts from psychology and cognitive engineering. The skill, rule and knowledge-based framework by Rasmussen (1983) was used to identify the main task types involved in the incidents. Error types such as post-completion error, prospective memory failure and data-entry error were used as part of the analysis. System usability and the occurrence of interruptions, distractions and multitasking were also adopted as important factors in the analysis. Results suggest that more than 60% of the incidents involved rule-based performance. Almost 50% of the rule-based incidents can be explained in terms of the errors types and factors identified. This analysis provides a starting point for practitioners to discuss aircraft maintenance incidents using theoretically grounded concepts.

Sign in / Sign up

Export Citation Format

Share Document