100 likes | 200 Vues
A Method for Evidence Based Quality Practice Engineering. Antti Hätinen 9.6.2009 TKK/SoberIT/ESPA Project. Background. Kitchenham: Evidence Based Software Engineering ESPA = Tekes funded research program in co-operation with Lappeenranta Technical University and TKK Four Subject Companies
E N D
A Method for Evidence Based Quality Practice Engineering Antti Hätinen 9.6.2009 TKK/SoberIT/ESPA Project
Background • Kitchenham: Evidence Based Software Engineering • ESPA = Tekes funded research program in co-operation with Lappeenranta Technical University and TKK • Four Subject Companies • 50-500 employees • Software Products for Civil Engineering, Electric Power Grid management etc.
RQ:How can a certain level of quality be reached? • Social methods • Experience factory
Evaluation of Social Search Methods • Quality Palette Analysis • Indicator Analysis • NMA • Brainstorming on quality indicators • With helping questions • How could the current situation be improved the most? • Which current practices can’t be obeyed? • What prevents/limits reaching the goal? • What could be done earlier?
Semantic Web Experience Base • Datamodel with 5 classes • Practice • Goal • Context • Result • Reference • Inference Rules (AI) • QPA Matrix plugin • ISO9162 Goal Closure • Web Interface • SPARQL+ queries
Semantic Web Experience Base • Input • Systematic literature reviews of all scientific results on SWE practices <- huge work • Vector of current Practices of a company • Vector of Goals with (current) and target values • Output • Optimal set of practices (=methodology) that has empirical evidence on reaching the desired quality level • E.g. Which (design) practices should be used to reach usability of 5 clicks/goal • SPI initiatives how the current operations can be improved by • Introducing new practices • Enhancing existing practices • Removing inefficient (anti)practices
Results • EBSE DB seems to be the best alternative for answering the RQ • While it is not widely available and fully populated, NMA/Brainstorming should be used for Software Process Improvement • Example of SPI Suggestions
Future Research • Database is a platform for Systematic Literature Reviews • After reviewing & populating the db with all existing scientific evidence, at least 100 SLR conference articles can be written on different practices • A paper on the advantages of using Semantic Web for EBSE • Compared to the related work: • http://evidencebasedse.com • https://bpch.dau.mil • Bayesian Network analysis for answering • How generizable are the results on students to industry context? • What are the ISO9162 metrics driving the quality most? • E.g. what is the best (most orthogonal) definition for usability?
Future Research • Ontology of SWE • Side-benefit for the academia for providing an exact vocabulary • Education on Software Engineering • The database can be used for • Reference index for all available research on each practice • Training material for teaching the current top 120 practices • Methodology research • How do the different sets of practices differ in performance? • Are there more optimal combinations = new methodologies? • How do Extreme Programming and Waterfall differ from each other?
Thank You! Questions?