IFIP Open Access Digital Library
Not a member yet
22614 research outputs found
Sort by
User Story Based Automated Test Case Generation Using NLP
No full textPart 1: Applications of AI/ML in Natural Language ProcessingInternational audienceThe progress of technology requires software systems to be of higher quality in order to meet the increasing complexity and frequency of changing needs. The present software development life cycle prioritizes the adjustment to evolving client requirements across the different stages of project development, facilitated by Continuous Integration and Continuous Deployment. The process produces a substantial volume of data that can serve as a valuable resource for automating test case production and reducing the need for manual intervention. This publication presents a suggested technique that utilizes natural language processing to automate the generation of test cases, hence minimizing the need for human involvement. The proposed approach has three phases: input-output categorization utilizing sentiment analysis, production of regular expressions, and generation of test cases. The main contribution of this article involves the classification of user keywords and the construction of test cases using them. The suggested model generates diverse outputs to create both positive and negative test cases. It has been tested with 700 user stories that have varying levels of abstraction in articulating the requirements
A Comparative Analysis on Various Machine Learning Methods for GAN Based Video Anomaly Detection
No full textPart 2: Applications of AI/ML in Image ProcessingInternational audienceIn recent years, surveillance has undergone tremendous change and developed into an essential instrument for maintaining security and keeping an eye on sensitive areas. This essay investigates the idea of what defines surveillance. It explores the crucial topic of anomaly detection, which is a vital component of contemporary surveillance systems. The expanding importance of the use of deep learning methods is highlighted in this paper’s discussion of current developments in surveillance technology. Through the processing of massive volumes of data, deep learning has transformed surveillance by allowing more precise and effective anomaly detection. Examining several kinds of deep learning techniques, their special qualities and uses are highlighted. This study concludes with an in-depth analysis of the monitoring, highlighting the role of deep learning in improving anomaly detection. It is an invaluable tool for researchers, professionals and decision makers interested in the development of surveillance technology and its use in different contexts
SVM-Based Skin Cancer Diagnosis for Malignant and Benign Tumor Distinction
No full textPart 2: Applications of AI/ML in Image ProcessingInternational audienceIn this exploration, we have successfully developed an SVM model that can enhance the process of skin cancer diagnosis. By utilizing advanced techniques in image analysis such as SVM, we have extracted unique features from dermatoscopic images of skin cancer, enabling us to detect distinctive dermoscopy patterns that indicate benign or malignant cancer tumors. The work aims to distinguish between benign and malignant growths, and we improved the accuracy of our diagnostic approach by using a support vector classifier and its capabilities. The work comprises the combination of dermoscopic image classification, machine learning, and, medical imaging, which together form a diversified diagnostic tool from various fields. After conducting attentive testing on different datasets to ensure the performance of our SVM model and the ease of use, satisfied outcomes occurred. The objective is to create a model i.e., a diagnostic tool to distinguish dermoscopic images and improve the early detection technology in skin cancer diagnosis to improve skin health of patients and raise awareness among the precautions to be taken
Analyzing Value-Sharing Methods in Energy Communities with Coalitional Game Theory
No full textInternational audienceWithin energy systems, collaboration has recently gained increased attention from academia and industry. However, the success of collaboration requires a fair value-sharing method based on the individual contributions. Coalitional Game Theory (CGT) offers a conceptual framework for analyzing projects where participants cooperate or collaborate to achieve favorable outcomes. The growing importance of collaboration in Energy Communities (ECs) suggests the development of CGT-based frameworks designed to understand the behavior and relationships among diverse players in a variety of operational and planning contexts. This work presents an overview of recent developments in CGT and their application in ECs, starting with a concise theoretical explanation of CGT, focusing value-sharing methods. It then examines recent applications of CGT in addressing operational and planning challenges within ECs. The paper concludes with a brief discussion on limitations, opportunities, and potential paths for further research in this field
LLM-Based Extraction of Contradictions from Patents
No full textPart 1: AI-Driven TRIZ and InnovationInternational audienceAlready since the 1950s TRIZ shows that patents and the technical contradictions they solve are an important source of inspiration for the development of innovative products. However, TRIZ is a heuristic based on a historic patent analysis and does not make use of the ever-increasing number of latest technological solutions in current patents. Because of the huge number of patents, their length, and, last but not least, their complexity there is a need for modern patent retrieval and patent analysis to go beyond keyword-oriented methods. Recent advances in patent retrieval and analysis mainly focus on dense vectors based on neural AI Transformer language models like Google BERT. They are, for example, used for dense retrieval, question answering or summarization and key concept extraction. A research focus within the methods for patent summarization and key concept extraction are generic inventive concepts respectively TRIZ concepts like problems, solutions, advantage of invention, parameters, and contradictions. Succeeding rule-based approaches, finetuned BERT-like language models for sentence-wise classification represent the state-of-the-art of inventive concept extraction. While they work comparatively well for basic concepts like problems or solutions, contradictions − as a more complex abstraction − remain a challenge for these models. Even PaTRIZ, the latest and complicated multi-stage approach to extract contradictions, delivers only mixed results. This paper goes one step further, as it presents a method to extract TRIZ contradictions from patent texts based on Prompt Engineering using a generative Large Language Model (LLM), namely OpenAI’s GPT-4. The existing annotated patent dataset “PaGAN” is used to demonstrate the LLM-capabilities for extracting TRIZ contradictions from the section “State-of-the-Art” of USPTO patents. Contradiction detection, sentence extraction, contradiction summarization, parameter extraction and assignment to the 39 abstract TRIZ engineering parameters are all performed in a single prompt using the LangChain framework. Our results show that “off-the-shelf” GPT-4 is a serious alternative to PaTRIZ. Comparing the text similarity of the GPT-4 extractions with the annotated sentences from PaGAN we reach a high F1-value of 0.93 using the BERTScore metric
On Opportunities and Challenges of Large Language Models and GPT for Problem Solving and TRIZ Education
No full textPart 1: AI-Driven TRIZ and InnovationInternational audienceThe advent of GPT has caused a real revolution in many application contexts. Even the TRIZ community has had to face up to this new technology, questioning the possible integrations with traditional paths and tools. Many problem-solving experts have for some time been proposing specific prompts based on the methodology’s tools such as functional analysis, reconstruction of cause-effect relationships, identification of Resources, 40 inventive principles, etc., in order to support the problem solver, or even replace him altogether, during the inventive process. The free generation of LLM content has been applied for very different purposes such as, for example, to contextualize general purpose heuristics in specific domains, or as a search engine to answer technical questions, to suggest creative ideas or improve the formulation and redefinition of a problem, or finally to find connections between different application contexts.This article proposes a critical analysis of the real effectiveness of these prompts according to the different needs of users.The analysis was carried out using a software application that was developed in-house and for which a testing phase was conducted on a variegated sample covering both the academic and industrial fields, with more experienced users and users who have been approaching TRIZ for less time
Partially Defined Logical Operators in Cause-Effect Models
No full textPart 3: Interdisciplinary and Cognitive Approaches in TRIZInternational audienceCause-effect models used in TRIZ to identify key disadvantages of the analyzed systems employ AND and (explicit or implicit) OR operators to indicate how the causes trigger the effects. These fully defined logical operators imply disadvantage elimination strategies, i.e., removing any of the AND-connected causes vs. removing all OR-connected causes. On the contrary, the partially defined logical operators incur uncertainty about the trigger conditions due to the output values being undefined for some input combinations. This paper proposes a method of handling such operators to support decisions regarding disadvantage elimination, which may interest TRIZ practitioners and researchers.The paper starts with recalling the basics of cause-effect analysis and introducing the notations for describing Boolean functions. The requirements and sample statistics concerning disadvantage descriptions are discussed in the second section, while the third section introduces partially defined operators. The following two sections present the proposed categorization of logical functions and the method for systematic selection of contributing causes to remove. The summary and ideas for further research are given in the last section of the paper
Systematic Prototyping Using TRIZ
No full textPart 3: Interdisciplinary and Cognitive Approaches in TRIZInternational audienceA typical TRIZ project aimed to develop or improve a system consists of the problem analysis, solution generation and solution substantiation phases. The last phase may vary in approach and scope, ranging from expert assessment to physical prototypes and digital twins. Prototyping activities are often iterative, and the results achieved with one prototype are typically used to develop another. The order of modifications may significantly affect the total cost, time, and effort of prototyping, so adequately managing this process seems to be a vital challenge.This paper proposes a systematic approach to prototyping navigation using TRIZ tools, such as Function Analysis and Value Analysis. If a new solution is to fail eventually during the substantiation, we would prefer it to fail fast and minimize the efforts required to obtain this result. Prototyping a successful solution, on the contrary, may benefit from a systematic approach by postponing the most significant investments to the latest stages when the design seems sufficiently reliable. The paper introduces the notions of prototyping attractiveness, uncertainty zone, and uncertainty time and provides simple guidelines for navigating the prototyping process, which are illustrated by a real-world example
Balancing High Social Welfare and Short Waiting Times: Determining a Reasonable Buyout Price in Auction-Based Restaurant Reservation Systems
No full textPart 4: Mechanism Design for Smart and Sustainable Supply ChainsInternational audienceThis study proposes an auction-based dynamic reservation system, the “synthetic scheduling auction” (SSA), to maximize social welfare through restaurant services. Each potential customer group submits 3-tuple bids to the restaurant, including the meal starting time, meal selection, and number of people in the group. After receiving these bids, the restaurant offers a buyout price to the group. To calculate the buyout price, the restaurant conducts an auction with synthetic customer agents who intend to book the restaurant within the reservation period. These agents are generated based on the restaurant’s past customer statistics data. The buyout price is the “expected nuisance fee” based on the money transfer function determined by the VCG mechanism applied for static auctions. By conducting an auction with synthetic customers, the restaurant determines the buyout price without waiting for actual future reservation applicants. Therefore, SSA achieves an increased social welfare than in the First-Come-First-Served (FCFS) reservation system, the current standard, and a decreased waiting time for each customer group compared with the static auction mechanism. Computer experiments were conducted on a fictitious restaurant to compare the social surplus and seat occupancy rates when the restaurant used the SSA and FCFS system. The results demonstrate that the SSA is superior in both metrics, that is, social surplus and seat occupancy rates
Inspection Planning Improvement Framework Based on the PDCA Cycle
No full textPart 1: Lean Thinking Models for Operational Excellence and Sustainability in the Industry 4.0 EraInternational audienceInspection Planning (IP) is critical to assure product quality during the Manufacturing Stage. IP takes place mainly during the stage of Process Development and is considered an inflexible static output resulting from the initial plan. However, revisions can become necessary according to the manufacturing system’s dynamic situation. Therefore, a flexible IP is needed, which can be achieved through a Continuous Improvement (CI) approach. Although the advantages of feedback mechanisms are renowned, the design of quality control loops is not common in practice. Specifically, there has been no application of the PDCA cycle aimed at performance improvement of IP. This paper proposes an IP improvement framework based on the PDCA cycle. The framework is applied in a case study in which IP changes were driven by a Kaizen Event. This paper provides new empirical evidence and extends the body of knowledge related to the CI of IP, showing that companies can stabilize and even increase their inspection performance by implementing improvement actions through the proposed framework. The case study describes alterations related to which quality characteristics to inspect and whereby, while suggesting that changes in inspection extent are more likely to happen during the ramp-up phase. It concludes that companies need to create and maintain structures and processes that enable a quick, effective response to internal and external change drivers through a well-structured approach. This framework provides an organized and easily understood management approach for the effective improvement of IP