スマートエスイーセミナー: ソフトウェア設計の技術的負債とモデリングフレームワーク

Seminar on Design Technical Debt and Machine Learning System Modeling Framework

設計のTechnical Debt、機械学習システムのモデリングフレームワーク

2024年3月5日(火)18:50-20:20 March 5th Tue 6:50pm-8:20pm JST 早稲田大学西早稲田キャンパス＆オンライン

ソフトウェアアーキテクチャの設計手法ADDや評価手法ATAM, SAAMなどを提案され、書籍『実践ソフトウェアアーキテクチャ』や書籍『Technical Debt in Practice: How to Find It and Fix It』の著者として著名なソフトウェアアーキテクチャならびに技術的負債の第一人者Rick Kazman教授（ハワイ大学、CMU/SEI、IEEE Computer Society理事）ならびに技術的負債やアーキテクチャ・ビッグデータ解析で著名なHong-Mei Chen教授（ハワイ大学）が来日されることを記念し、複雑・不確実な開発運用時代に重要性を増すソフトウェア設計とモデリングについてセミナーを開催します。具体的にはChen教授から技術的負債の概念を解説いただき、続いてKazman教授から設計における技術的負債のプロジェクトへの影響や特定ならびに修正のアプローチについて解説いただきます。さらに、JST未来社会eAIプロジェクトにおける早稲田大学チーム（チーム代表: 鷲崎）における機械学習システムの設計を含むモデリングフレームワークの取り組みを解説します。なおすべて英語による実施となります。ぜひご参加下さい。

Prof. Rick Kazman (University of Hawaii and CMU/SEI), the author of the books "Software Architecture in Practice" and "Technical Debt in Practice: How to Find It and Fix It" and a leading expert on software architecture, and Prof. Hong-Mei Chen (University of Hawaii), a well-known expert on technical debt, architecture and big data analysis, will visit Japan to give a seminar on software design and modeling, which are becoming increasingly important in the era of complex and uncertain development operations. Professor Chen will explain the concept of technical debt, followed by Professor Kazman's presentation on the impact of technical debt in design on projects and approaches to identify and correct it. In addition, the Waseda University ML framework team led by Prof. Washizaki will explain a modeling framework that includes the design of machine learning systems. All sessions will be conducted in English. We look forward to seeing you there.

主催

• スマートエスイー

協賛・協力

• 早稲田大学グローバルソフトウェアエンジニアリング研究所

概要

• 日時: 2024年3月5日（火）18:50-20:20
• 場所: 早稲田大学西早稲田キャンパス 63号館 5階 0506会議室、ならびに、オンラインZoom
• 参加申込: 本connpassページからお申し込みください

プログラム（個々の時間や内容は変更の可能性があります）

18:50-18:55 オープニング Opening

18:55-19:10 Introduction on Technical Debt（技術的負債とは）

Prof. Hong-Mei Chen (University of Hawaii at Manoa)

Biography: Hong-Mei Chen is a Professor of Information Technology Management at the Shidler College of Business of the University of Hawaii at Manoa. She formerly served as Associate Dean for the College and was the Founder and Director of the Advanced Information Management Solutions (AIMS) Lab. She has won prestigious teaching excellence awards at the college, community, and university levels.

She conducts cross-disciplinary empirical research on information systems design and development including technology adoption, frequently working with many executives of Fortune 100 companies. She has obtained multi-million-dollar grants and directed several large-scale, multi-institution research projects. She directed and implemented the image database system for the DARPA MISSION project utilizing NASA experimental satellites. She was the director for the National Data Center for the $34 million Electrical Vehicle development program. She serves on US NSF (National Science Foundation) review panels and a NSF large grant (>$25 million) management team. She has published in prestigious MIS and Software Engineering journals in areas such as social debt, big data engineering and management, software architecture, innovation-driven system design methods, green information systems, cybersecurity, crowd-sourced systems, social CRM (customer relationship management), business-IT alignment, service engineering, distributed database, and AI.

19:10-19:50 Finding and Fixing Design Debt（設計における技術的負債の発見と修正）

Prof. Rick Kazman (University of Hawaii, CMU/SEI)

In this talk I will discuss a common and pernicious form of technical debt--called design debt, or architecture debt. I will briefly present the theoretical foundation behind this form of debt and present a broad set of evidence demonstrating its dramatic effects on project outcomes. That is the bad news. The good news is that we can automatically pinpoint the causes and scope of such debt. I will describe how we can automatically locate it, measure it, and create the business case for removing it. Finally, I will explain how we can remove--pay down--this debt via refactoring. I will also sketch some of my experiences doing all of this in real-world projects, along with the outcomes.

Biography: Rick Kazman is the Danny and Elsa Lui Distinguished Professor of Information Technology Management at the University of Hawaii and a Visiting Researcher at the Software Engineering Institute of Carnegie Mellon University. His primary research interests are software architecture, design and analysis tools, software visualization, and technical debt. Kazman has been involved in the creation of several highly influential methods and tools for architecture analysis, including the ATAM (Architecture Tradeoff Analysis Method) and the Titan and DV8 tools. He is the author of over 250 publications, co-author of three patents and eight books, including Software Architecture in Practice, Technical Debt: How to Find It and Fix It, Designing Software Architectures: A Practical Approach, Evaluating Software Architectures: Methods and Case Studies, and Ultra-Large-Scale Systems: The Software Challenge of the Future. His research methods and tools have been adopted by many Fortune 1000 companies and has been cited over 29,000 times, according to Google Scholar. He is currently a member of the IEEE Computer Society’s Board of Governors, an Associate Editor for IEEE Transactions on Software Engineering, and a member of the ICSE Steering Committee.

19:50-20:10 Multi-view modeling framework for reliable machine learning systems（高信頼機械学習システムのためのマルチビュー・モデリングフレームワーク）

Jati H. Husen, Jomphon Runpakprakun, Hironori Washizaki (Waseda University)

To address the analysis between the experimental nature of machine learning and the deterministic side of traditional software engineering requires specific approaches. To address the challenge, the eAI project's framework team is developing a Multi-view Modeling Framework for ML Systems (M3S) as a model-based framework that facilitates consistent and comprehensive analysis of ML systems. The analysis includes integrating the modeling environment and the ML pipelines to facilitate the highly experimental characteristics of ML models, in which a series of training and evaluations are conducted with different solutions and configura tions to identify an ML model version that satisfies all requirements. These integrations are based on a cohesive metamodel to ensure analysis consistency. The underlying approach of M3S supports a reliable and comprehensive analysis of the ML system while ensuring tight synchronization with the implementation. This synchronization provides the base for a feedback loop between the analysis and the implementation at both the ML model and the overall system levels.

20:10-20:20 議論とクロージング Closing