Connection and convenience come at a cost, particularly in the Internet of Things (IoT), where smart devices such as bulbs and surveillance cameras widen the attack surface for malicious actors. Intrusion Detection Systems (IDSs) are a critical line of defense, but their effectiveness must be balanced with strict resource constraints across edge, fog, and cloud layers. In this work, we propose lightweight machine learning (ML) and deep learning (DL) models for deployment across the edge–fog–cloud continuum, evaluating eight models on the RT-IoT2022 dataset. Using Optuna for hyperparameter tuning, we find that the ensemble models perform best, with LightGBM achieving the highest macro-F1 score (98.07%). However, a compact 28.8 KB Fully Connected Neural Network (FCNN) offers a favorable trade-off, attaining a macro-F1 of 94.80% with consistent sub-millisecond inference and relatively low power usage across all devices. Notably, while neural networks may not always match ensemble models in classification performance, they exhibit comparatively lower variability in inference times, making them attractive where predictability and throughput stability are crucial. Our statistical testing confirms the superiority of ensemble models while quantifying the gaps in performance. We further analyze the deployment across diverse hardware and compare ONNX runtimes against native C/C++ code generation to illustrate the convenience–performance trade-offs. We note that while quantization preserves accuracy, it can increase latency or power consumption for already-small models, cautioning against its blanket adoption. Finally, we release a reproducible end-to-end pipeline and provide practical deployment guidelines to support model and device selection under heterogeneous IoT constraints.

Since the advent of COVID-19, educational institutions worldwide have increasingly adopted online learning where challenges in maintaining student engagement have been observed. Consequently, innovative solutions for monitoring and improving engagement in virtual environments have been explored. This paper presents the design and implementation of an affordable, efficient device equipped with a camera that infers student engagement in real time using Deep Learning techniques. The system utilizes a convolutional neural network to analyze visual data and classify student behavior into three categories; focusing on paper, focusing on the screen, and being distracted. To ensure compatibility with resource-constrained edge platforms, the trained model was quantized using TensorFlow Lite (TFLite), enabling efficient deployment across a variety of devices, ranging from low-power microcontrollers like the Sony Spresence and OpenMV cam to more powerful GPU and TPU-based edge platforms such as NVIDIA Jetson Orin and Google Coral. We also established a baseline on the Raspberry Pi 5, optimizing for performance and resource constraints. A key focus of the work presented here was safeguarding student privacy. All data processing occurs locally on the device, ensuring that no sensitive information is transmitted externally. The quantized model enables real-time performance without compromising privacy, as all engagement data is securely analyzed on the edge device; only the student's engagement state is transmitted across potentially insecure networks, while the sensitive video feed remains confined to the student's device. A comprehensive comparison between the quantized, non-quantized, and quantization-aware models was conducted to assess the impact of quantization on accuracy and inference speed. In addition, trade-offs between power consumption and performance across different devices were evaluated. These findings demonstrate a balanced approach to achieving computational efficiency while ensuring real-time student engagement monitoring. By leveraging these cost-effective technologies, our approach offers an accessible and scalable realtime student engagement monitoring tool and addresses privacy challenges posed by remote education, even beyond the pandemic.

Measuring student engagement can have a significant impact on student learning and can help assess and improve pedagogical interventions. Deep learning has been used successfully on student videos to label student engagement. The long-term goal of the research presented hear is to build a small device that uses a tiny camera to locally infer student engagement. This paper evaluates various backbone neural network architectures and shows that it is possible to achieve state of the art (SOTA) performance of over 95% accuracy and 95.93 Fl-score on a student engagement data set collected during online math sessions. In addition the trained models were deployed on edge devices and demonstrated that it was possible to achieve a performance of over 16 frames per second using the Nvidia Nano edge GPU.

Generating content for large-scale reading comprehension assessments, particularly stories, poses significant challenges including money, time, and effort. To serve as effective assessment tools, these comprehension stories must adhere to specific criteria defined in early grade reading assessment standards dictating narrative structure, character types, readability levels, and other elements. Recently Natural Language Processing (NLP) techniques, mainly Large Language Models (LLMs), have been used to automate the story generation process. One key challenge is ensuring diversity across the many generated stories. For example, the Early reading standards such as Early Grade Reading Assessment (EGRA) requires comprehension stories to be similar in difficulty level but narratively different across multiple implementations to ensure fairness. This paper proposes a solution to increase diversity across stories by employing GPT-4 to generate children’s reading comprehension stories mediated by a database of classic tales. By leveraging existing narratives, this method drastically reduces the resources required for content creation while ensuring alignment with the EGRA criteria. We present a systematic framework for selecting, adapting, and evaluating the stories, aiming to streamline the content generation process. The generated stories are additionally evaluated using well-defined text metrics and by a human evaluator for reliability. Our findings underscore the potential of integrating GPT-4 with classic tales to optimize resources and enhance scalability in literacy assessment practices, offering practical implications for educators and policymakers in early grade learning.

Online learning platforms are generating an enormous amount of data that lends itself to unsupervised learning. This paper presents a case study where assessment data from two online platforms was used to cluster students into similar groups. The long-term objective of this research is to incorporate the clustering information into the personalization mechanisms. K-means was used to cluster students for 10 Skills. K-means was able to create a small number of clusters with reasonable internal validity with an average silhouette width of 0.32 (sd=0.05). The clusters were non-trivial as gender, school or class could not explain the clustering with an average Adjusted Rand Index (ARI) of 0.049 (sd=0.03). Most importantly, only a small subset (18%) of attempted questions could be used to explain accurately (Average F1-measure = 89.43) why the students were grouped into clusters. These keystone questions can be used to further enhance the personalization mechanisms.

The Internet of Tangible Things (IoTT) provides new opportunities for building technology-enhanced learning systems that wirelessly connect tangible sensor data to remote learning services. This paper presents the design and implementation of an IoTT voice driven learning platform that uses Amazon Alexa along with sensor-enabled wireless dice to build tangible and interactive math games for kindergarten and primary school children. The wireless dice used sensors in a 9-axis IMU to accurately detect the face each die falls on and sent this information to Amazon Web Services (AWS) using the MQTT protocol using AWS-IoT service to enact arithmetic learning scenarios. An initial evaluation shows that most children from ages 5 to 9 enjoyed playing with this system. Half the children liked everything about the game, and they all liked the concept. This game is a first step towards a new genre of educational learning technology called Internet of Tangible Learning Things (IoTLT).

Smart watches offer a unique opportunity to enact ubiquitous learning scenarios. The various built-in sensors in these watches provide contextual information that can be directly incorporated into ubiquitous learning designs. This paper presents the design and implementation of a wearable assessment platform based on the Samsung Gear S2 Smart Watch. Implemented using the IoT MQTT protocol, this platform allows the system to automatically subscribe to and publish questions to students based on their current context. For example, students will get notified on their smartwatch with a question related to their location or situation derived from the smartwatch sensor data. The system uses the PONTE MQTT server and is integrated with Google's Classroom.

Classroom observation is an important tool for improving quality of education. Observations allow teacher trainers to monitor, assess, and provide feedback to teachers about their teaching techniques and class management practices. The current approach to class observation consists of an observer visiting a classroom once or twice a year. This is both costly in terms of time and resources, insufficient in terms of capturing the true state of the classroom, and in providing accurate feedback. This paper presents the design and implementation of a mobile-based system that automates the class observation process using data mining techniques. The system allows teachers to record audio snapshots during the class session, and receive Stallings-type analysis results for their class on a daily basis.

Inter-class and within-class grouping has been used extensively to deal with student diversity. This paper presents a method that uses Rasch Modelling and K-means to automatically create ability-based groups or prototypes of students that can be utilized to provide customized lesson plans and advice to teachers. A second-order clustering is used to create groups of teachers to help deliver more effective continuous professional development for teachers. A case study of applying the method to a deployment of Early Grade Reading Assessment (EGRA) in two rural and semi-rural Districts in Pakistan is presented. Data from 99 schools and 717 students is considered. The results are that approach produces developmentally interesting prototypes or groups of students in addition to creating key insights on how to organize professional development activities for teachers.

The availability of built-in sensors in mobile phones has enabled a host of innovative applications. One class of applications deals with detecting a user's emotions. Previous applications have primarily relied on recording and displaying self-reported emotions. This paper presents an intelligent emotion detection system for mobile phones implemented as a smart keyboard. The smart keyboard independently infers a user's emotional state using machine learning techniques. The system uses accelerometer readings and various aspect of typing behavior like speed, number of backspaces, and time delay between letters to train a classifier to predict emotions. Naïve Bayes, J48, IBK, Multi-response linear regression and SVM were evaluated and J48 was found to be the best classifier with over 90% accuracy and precision. In addition to providing emotive feedback to individual users, the system also uses geo-tagged data to collect and display emotional states of regions or countries through a website.

Educational analytics have a great potential to personalize learning processes in the developing world. Most schools in the developing world follow fixed curricula in their schools with little room for personalization. Infrastructure and cost constraints make the use of 1-1 computer-based tutors to personalize student learning unrealistic in these contexts. However, pervasive presence of mobile and tablet devices has made it possible to collect student assessment data, and to personalize decision-making for teachers and educational administrators. This paper presents an educational analytics method based on Knowledge Space Theory (KST) and KMeans to group students based on their zone of proximal development (ZPD). The paper also presents a case study applying this technique to support educational administrators and teachers in semi-rural schools in a developing country.

While cornerstone and capstone courses teach ”design” at the beginning and end of many computer engineering programs, a design gap is left in the sophomore and junior years. This paper describes the development and experiences of teaching five semesters of a specialized design course for a computer engineering program taught in the junior year. The course includes topics in software design patterns, multi-threaded applications, rudimentary user-interface design, embedded programming including sensor building and processing, and Zigbee-based wireless communications. The course is 2 semester credits (1-2-3), and includes a comprehensive project that requires students to build a tangible game or a game-like entertainment system. One key component of the course is the use of an open-source hardware platform called Arduino. The results are that students enjoy the course and are able to build integrated systems that incorporate design patterns, multithreading, embedded programming and wireless communication in an effective manner within one semester.

This article presents the results of a design workshop that developed scenarios of learning in an imaginary "smart city": Villard-de-Lans (Vercors, French Alps) by exploring suitable methodological approaches to the Smart City Learning Design. The aim of this case study was to propose "glocal" solutions intended to balance the global trends and the expectations of the community of reference (local elements), while distinguishing between access to information and participation in informal learning processes located in a smart city.

Ubiquitous learning technologies will play a key role in smart learning cities. However, despite the hype, the degree of adoption of ubiquitous learning systems, even in controlled academic environments, has perhaps not reached its full potential. Consequently, there is a need to understand the adoption behavior of key stakeholders. This paper presents a first step in this direction by asking researchers who have published recently in the area of 'ubiquitous learning' about which adoption factors the future users of their technology would prefer. The results indicate that for students, teachers and administrators, the researchers believe that social influence was the key variable affecting adoption. However, in addition to social influence, for teachers and administrators, performance expectation and effort expectation, respectively, were also considered important. Surprisingly, researchers did not agree on performance expectation as a key driver of the adoption of ubiquitous learning technologies.

With the advent of cheap and readily available body-worn sensors, new electronic pedagogical approaches are emerging. This paper describes the design and development of a prototype electronic training system for the game of Cricket. The system is designed to provide real-time feedback to players and coaches to help improve technique and performance in a game. The system uses a watch with a built-in accelerometer that transmits real-time 3-axis acceleration data from a player using a proprietary wireless protocol. Machine learning techniques are used to analyze this data to provide real-time critique to the player. Player's performance data is also recorded in a Learning Management System (LMS) and a coach or the player can view various reports related to their past performance. The system has been implemented using Dynamic Time Warping (DTW) and shows over 93% accuracy.

This paper presents the design, deployment and evaluation of a blended learning intervention that used Khan Academy videos to teach Grade IV and Grade V Mathematics to children in a developing country. A unique aspect of this learning intervention was the use of a self-contained, satellite-enabled learning van that visited a school once every two weeks to provide learning access to children and teachers in remote areas. Pedagogy included teacher mentoring and structured feedback to students and teachers based on online formative and summative assessments using Android tablets. Six pairs of control and treatment groups were used to evaluate this intervention. Based on results of standardized test instruments, the treatment groups outperformed the control groups in five out of six cases, observed effects were from 0.86 to 2.5 indicating that the approach worked well.

Pleasure is a key component of any game design. This paper presents a pleasure framework based on sixteen dimensions of pleasure including creation, discovery, sensation, danger and sympathy. The utility of the pleasure framework is shown by applying it to a tangible game-based learning system designed to teach university students about WiMAX signal propagation. The analysis reveals that while eleven pleasures were considered in the design of this game, the game could be improved along each dimension. Further, five pleasure dimensions were not explicitly considered at all and included important variables like fantasy, discovery and subversion.

Remote laboratories are traditionally being used to provide access and to augment and enhance existing traditional pedagogical models in an online environment. This paper proposes a generalized methodology to build a class of learning games for remote labs. The methodology relies on successive rounds of problem posing and problem solving. A case study in chemical engineering is used to illustrate the methodology.

Many children's games have a universal, cross-cultural appeal and have been played for hundreds of years, suggesting a developmental need for them. This paper presents a framework for leveraging this appeal and longevity by developing cognitive games based on popular playground games. This framework employs technology to maintain the physicality and embodied components of the game. The Prête-à-apprendre (PAP+) toolkit is used to develop an e-Textiles game for teaching concepts of estimations and number sense in Mathematics. The game consists of "hoodies" and "barrels" that communicate with each other using a low-power wireless network (Zigbee). Each "hoodie" contains an LED display to convey constraints to children while each barrel contains and Radio Frequency Identifier (RFID) reader to identify children. The game can be deployed in structured or unstructured environments. An evaluation of the pedagogical design of the game based on various intrinsic and extrinsic motivation criteria is also presented.

A Clicker or a wireless student response system is an example of a high density simple response ubiquitous system that processes real-time simple responses from a large number of students in a controlled environment like a classroom. Currently such systems are expensive and use proprietary hardware which makes it difficult to extend them for novel learning scenarios. This paper presents the design and implementation of a hybrid low-cost open-source hardware platform called Info Coral that uses the 1-Wire wired protocol in conjunction with Bluetooth or a Zigbee wireless interface. The hardware also employs the open-source Arduino software for ease of programming and customization. A Case study of applying this platform to build a low-cost clicker system for a K-12 classroom is also presented.

Success of Wikipedia has opened a number of possibilities for crowd sourcing knowledge. However, not all crowd sourcing initiatives are successful. This paper presents a preliminary study to determine if teachers in a developing country are able to create quality multiple-choice questions for primary school students. In addition, an adoption model is developed and evaluated to ascertain if the teachers would actually contribute to such a Wiki. Results are that a reasonable number of teachers are able to formulate quality questions in Science and English and that there is a strong intention to use such a system. However, there is no obvious relationship between the intention to use and an ability to pose good assessments.

Embodied Ubiquitous Learning Games (UELG) represent a new genre of technology-enhanced education in which learners interact with augmented physical environment to learn. The embodiment in such games comes from instrumented clothes based on e-Textiles incorporating sensors, actuators and low-power wireless modules. Designing and building such ubiquitous systems requires a complex interplay of conventional and embedded hardware and software co-design that includes novel elements like e-Textiles. In order to reduce the cost and shorten the development life-cycle this paper proposes the use of Software Product Lines (SPL). Towards this end, this paper borrows ideas from Architecture to present and apply a framework for constructing feature models for a Software Product Line for ULEG. The framework explicitly uses different types of features related to pedagogy, technology and domain of learning. A family of UELG based on a combination of problem posing and problem solving is used as an example to explain various components of the framework.

A number of adaptive mechanisms for asking questions have been proposed. This paper presents a novel adaptive method for selecting questions from various pools of questions. The method uses a state-diagram to represent various question pools where transitions between the states are dynamically modified to generate the optimal pool of questions for the next question. C4.5 is used to close the loop on adaptive selection of questions. A case study showing how the method was implemented for an online environmental game is presented and evaluated.

This paper describes the design and implementation of a ubiquitous learning system called prête-à-apprendre+. The system is based on e-textile components and consists of shirts and fabric pads that communicate through the Zigbee wireless protocol. A simple tag game based on this ubiquitous platform is described.

With global warming, environmental awareness has taken a front-stage. Awareness of environment is particularly important for the younger generation. Game-based learning represents a perfect platform for reaching teenager on this important topic. This paper presents the design, implementation and evaluation of a game-based learning system designed using the Neverwinter Nights role-playing shell. The game uses the narrative game design pattern. Initial evaluation of the system shows that the children enjoyed the game and thought that it was an effective and natural medium for conveying environmental awareness.

Learning activity based standards like the IMS Learning Design (IMS-LD) have recently emerged. Typical implementations of this standard rely on client-server architectures. This is problematic for resource limited mobile devices that may not have reliable Internet access. This paper presents the design and implementation of a system that takes a sub-set of the IMS LD design specification involving multiple roles and creates native code for mobile devices that collectively run the learning design using SMS messages for synchronization. An implementation targeting the Googlepsilas Android mobile platform is shown.

While software design patterns have been proposed for a wide variety of software domains, surprisingly little has been done to formalize an online assessment methodology. In other words, how does one assess a software engineerpsilas learning of a particular design pattern? This paper presents a framework and a tool motivated by Bloompsilas Taxonomy that extends an existing OWL ontology for design patterns in combination with assessment-generation heuristics to generate online assessments for software engineers. The resulting assessments are expressed in XML using the IMS-QTI standard. The paper also presents example assessments for GOF design patterns.

Ad-hoc sensor networks provide a cheap and scalable technology for constructing pervasive learning assessment systems that are embedded in physical environments. This paper proposes an extension to the QTI assessment standard that supports localized sensor data from sensor networks by incorporating the Sensor-ML notation. This extension can lead to a new class of pervasive learning environments where learning is enhanced by interacting with a large number of ad-hoc wireless nodes in a physical environment. Each wireless node can present various questions related to the current physical context. An architecture and a prototype implementation using the Jennic Zigbee toolkit is also presented.

It is important for software engineers to have a correct understanding of the software process they are following. Recently, standards for process modeling like OMG's Software Process Engineering Meta-Model (SPEM) and associated tools like the Eclipse Process Framework (EPF) have emerged. These standards allow a fine-grained description of a process to be conveyed to a software engineer in the form of specialized websites. However, a mechanism to determine how well a software engineer really understands a particular software process is still lacking. This paper presents a competency framework for software process understanding An ontology and a system that automatically generates such assessments for the Scrum software engineering process is also described. Protégé is used to construct the ontology while Jena 2 and Velocity are used to generate IMS QTI-based assessments that are automatically converted to Adobe Flash Lite format. The assessments are rendered over the internet and the results are directly stored in the Moodle learning management system.

Like other organization-wide compliance regimes, anti-money laundering training is an ideal candidate for use of e-learning technologies. The post 9/11 environment and recent penalties borne by banks engaged in money laundering activities further outline the need for these technologies. This paper presents a study on how e-learning technologies are being used in UAE banks. In specific, this study isolates the factors that may have impeded the introduction of such technologies. A host of factors including innovation diffusion and organizational and environmental characteristics do not explain why certain banks have not adopted this technology for anti-money laundering training. However, e-learning readiness, the perceived complexity and organizational slack seem to be the major impediments to adoption.

Many toolkits for extending the physical metaphors for learning have been proposed over the years. This paper presents the design and implementation of a toolkit for building tangible learning applications based on the IEEE 802.15.14 based sensor wireless networks. This toolkit provides a connection-less and a generic object-oriented programmable environment. A sample application showing how young children can browse the Internet using tangible components is also presented.

With the advent of m-learning, efficient rendition of XML-based assessments such as the QTIv2.1 is increasingly important. However, XML processing requires significant resources on resource-limited mobile devices. This paper presents an approach and architecture that bypass this problem by automatically generating self-contained Flash-Lite assessments from QTIv2.1 packages. A specialized object model of the QTIv2.1 is used in conjunction with open-source tools to generate assessments that are approximately five times smaller than those manually generated using macromedia learning interactions.

Soloman-Felder learning style index has been applied extensively in programming classes to ascertain the learning styles of students. This paper presents an approach showing how learning styles of students can be used to evaluate pedagogical resources. In specific, learning style can be used to help determine an appropriate textbook and an appropriate mixture of additional pedagogical devices such as virtual labs or 'clickers'. An example from a first undergraduate programming course is used to illustrate the approach.

Search engines like Google play a critical role in life-long learning. However, the query capabilities of such engines remain simple and often yield a result set that is too large. In addition, search engines like Google rely on page ranking algorithm that represents the "collective consciousness" of millions of users. Learning about specifics often involves context. This paper shows how mind maps can be used as a contextual mechanism to specify what needs to be learned and to filter and retrieve the relevant sources of learning from the Internet. In specific, a concept of "alignment" is introduced that filters only those information sources that are globally aligned to the mind map. The filter is implemented as a combinatorial optimization algorithm using Simulated Annealing

Much is understood about success factors, issues, and lessons learned in implementing e-learning in the context of higher education. A consensus is emerging that suggests that each institution is unique and hence requires a highly customized e-learning strategy. This paper presents a strategy formulation process that takes a community building view towards strategy formulation. In specific, the process uses the future search conference and the Cademenos model of growth of e-learning within higher education. An application and evaluation of the strategy formulation process at an American University is presented

Digital storytelling has been proposed for reflective practice and hence to improve learning in communities of learners. However, it is not clear if storytelling can have an impact on specific learning outcomes in higher education. In this paper we present a case study of digital storytelling in a civil engineering laboratory and assess the impact of storytelling on course outcomes. The primary result of the study is that although students perceive that they are learning better, storytelling does not seem to have any direct correlation with performance on learning outcomes in this higher educational context.

Many P2P frameworks for e-learning and knowledge management are being proposed. Most of these frameworks provide generic interfaces to support multiple information providers. This results in an implementation overhead. In this paper we present a lightweight P2P e-learning architecture which uses a fast native XML database to allow basic query, retrieval and download services using SCORM metadata. Like EDUTELLA, this environment is built using the JXTA framework. We also provide experimental data for this environment with respect to scalability in terms of the number of learning objects per peer, the number of concurrent queries, the number of peers and the size of the response set.

Semantically rich domains require operative knowledge to solve complex problems in real-world settings. These domains provide an ideal environment for developing authentic constructivist learning environments. In this paper we present a framework for development of authentic learning environments for such domains.

With increasingly short life-cycles, hand-crafting authentic constructivist environments for learning becomes increasingly unfeasible. However, as the complexity of designed systems and devices increases, so does the role of simulations to support design. This paper presents an approach and an example of how naturally occurring simulations in the design process can be used to automatically generate authentic constructivist learning environments.

Requirement analysis is one of the most important phases of application development. Numerous techniques for object-oriented analysis have been proposed. Automation in these methodologies, however, consist mostly of editors that support notation. Automating parts of the requirements analysis strategy will lead to more standardized specifications, will increase productivity and may reduce the number of bugs at this crucial stage. Unlike notation which is static and can be easily formalized, the strategy of requirements analysis is dynamic making its formalization difficult. Further, it is not even clear which parts of the strategy can be formalized. In this paper the authors present a framework for automating strategies of object-oriented analysis and describe a prototype system, CanVVaS, (Capture, Validation, Verification, and Synthesis) they have constructed to explore these issues.>

Although a host of testing methods have been proposed for expert systems, little work has been done to compare effectiveness of these methods. The authors present an analysis of various parameters that govern the effectiveness of output based partition testing strategies for heuristic classification expert systems.

Three object-oriented analysis techniques: Shlaer & Mellor, Coad & Yourdon, and Document-Driven Analysis were evaluated. Criteria for comparison of analysis techniques were developed. These criteria state that an analysis technique should (1) cover data, transformation, and control perspectives of the problem, (2) address large problems through partitioning, bounding the problem, and supporting prototypes and incremental analysis, (3) provide approaches for verification and validation, and (4) support configuration control. A case study in a software engineering class was conducted using the three OOA techniques. Through the case study we found that (1) identifying objects is affected by notation and flexibility of object models, (2) visualizing global data, transformations, and control are necessary to capture end-user’s processes, (3) emphasizing hierarchical structures and third normal form divert the analyst’s attention from analysis to implementation, (4) an explicit representation of relationships is necessary, (5) peer review is useful in improving selection of objects and methods to support end-user processes, and (6) supporting a standard document from OOA products is not easy.

The authors describe a life-cycle approach for testing expert systems. They comment on the difficulties associated with testing of an expert system. Pragmatic testing methods of conventional software engineering are proposed as a solution to these problems. The application of these techniques is illustrated through an extended example for the expert system, MAPS, which diagnoses faults in a bipolar transistor induced to process defects in a VLSI manufacturing environment.>

Effective testing (validation, verification and evaluation) for expert systems is becoming important. A comparative evaluation of black-box, white-box, consistency, and completeness testing methods is performed based on the criteria of effectiveness, robustness, and cost. Testing methods are evaluated using life-cycle mutation testing on a VLSI manufacturing diagnostic expert system. The results indicate that dynamic-flow testing can be used to catch faults in all phases of software development.>

Object-oriented analysis is the activity of understanding a problem domain and developing a specification for it within an object-oriented paradigm. In this paper we propose object-oriented analysis can be viewed as a design task that consists of analysis, synthesis and internal and external completeness and consistency checking. We postulate a blackboard based design framework and propose several alternative theories within this framework. Data from an expert analyst is collected and analyzed for relative explanatory power of the alternative theories. Basic findings are that viewed as a design activity, object-oriented analysis is neither top-down nor completely opportunistic in nature, Rather, (1) analysis (the process of finding objects and processes) and synthesis (constructing the object model and state-transition diagrams) are complementary activities while internal completeness and consistency checking is coupled with synthesis and (2) construction of objects and processes are cyclic and complementary processes. The findings are compared with both descriptive and normative research in automated software design.

In this panel session, the following topics are discussed: artificial intelligence in business; artificial intelligence in multimedia; neural networks as a tool for artificial intelligence: software engineering for knowledge-based systems: and artificial intelligence as a solution for software engineering.>

Summary form only given. Discusses the applicability of the object-oriented paradigm to the stage of problem specification. The authors define some criteria that can be used to assess this applicability and show how the object-oriented approach measures up based on these criteria.>

Issues that need to be addressed concerning the application of expert-system technology to software engineering are discussed. It is concluded that one should approach expert system use in software engineering with great caution, realizing that software engineering problems are ill-structured. A strong relationship abides between the specificity of software development methodology and the applicability of expert technology. If the methodology is to be very specific, no probable need for an expert system exists due to the availability of stronger methods. If the methodology is too weak, on the other hand, the problem is likely to be too ill-defined for expert system solutions.>

We would like robots to recognize and handle situations that do not conform with normal operating conditions. We want to be able to do this without having to consider explicitly errors caused by missing or defective parts, or by malfunctioning. To this end we present the detailed design of a system in which the controller of the robot takes advantage of large knowledge bases to ensure proper execution of the robot task. Real time considerations played a large role in our design.