1. University of Southern California Center for Software Engineering C S E USC August 2001©USC-CSE1 CeBASE Experience Base (eBASE) -Shared Vision Barry Boehm, Vic Basili, and CeBASE Team ISERN Presentation August 2001

2. University of Southern California Center for Software Engineering C S E USC August 2001©USC-CSE2 System Capability Description Provide better access to empirical data and models –And better understanding of their relevance Provide Web portal to categorized software engineering data, models, and methods –More extensive –Better categorized –Easier to use and contribute to

3. University of Southern California Center for Software Engineering C S E USC Populate KB in COTS-Based Systems (CBS) Defect Reduction (DR) areas Collaborative empirical research in CBS, DR Expand to other collaborators, areas, best practices Construct knowledge- based (KB) framework Integrate UMD, USC data, models Develop CeBASE portal Outreach and technology transition activities: Contact, awareness, understanding, trial use, limited adoption, institutionalization Initial Empirical KB in CBS, DR areas Growing KB of empirical best practices models in CBS, DR areas Growing community of CeBASE users, empirical researchers; KB content Much more predictable development of both rapid and robust software Critical-mass community of empirical SW researchers; scientific understanding of software phenomenology Wide, easy access to CeBASE KB : Empirical best practices, models, data Empirical research tools and techniques Empirical KB content KB form Compatible data, models Access facilitation New results more users Initial KB, empirical techniques Growing KB, empirical techniques Expanded results CeBASE Results Chain CeBASE usage cost-benefit justification Continued funding

4. University of Southern California Center for Software Engineering C S E USC August 2001©USC-CSE4 Key Stakeholders and Win Conditions Users: practitioners, empirical researchers –Easy, rapid, dependable, low-cost operation –Search, navigate, download, contribute –Metadata for understanding data context Customer: NSF –Scientific and practical value Developers and Maintainers: CeBASE team –Strong IOC by Spring 2002; easy to run and evolve Interoperators: International, DACS, others –Compatible data definitions –Data cross-link and aggregation

5. University of Southern California Center for Software Engineering C S E USC August 2001©USC-CSE5 System Boundary and Environment  Web portal  Search/navigation engine  Feedback channels  eDBA functions  eBASE content  FAQ, Chat/Discussion, Email, etc. ... Users Administrators Interoperators Middleware for portal, DB, archive management

6. University of Southern California Center for Software Engineering C S E USC August 2001©USC-CSE6 eBASE Content Empirical Data and Models: Initial focus on defect reduction –Organized around opportunity tree –Including metadata, references, crosslinks, implications for project practice; comments and contributions area –Basic search and navigation capabilities –Ability to enter user-project metadata, do similarity-search for most relevant results Project Archives: Initially MBASE, others TBD –Organized around project guidelines (MBASE, Q-Labs, BORE, …) –Including source artifacts and project data (effort, schedule, defects, customer satisfaction, benefits realized) Empirical Research Methods –E.g. experimenting, data mining, model building –Usage guidelines and experience reports, references, cross links; comments and contributions area –Common experiments eBASE access and usage data …

7. University of Southern California Center for Software Engineering C S E USC August 2001©USC-CSE7 Major Project Constraints Prototype eBASE by Aug. 20, 2001 Initial Operational Capability by May 5, 2002 –USC student team project Incremental approach; opportunistic priorities –COTS, defect reduction data –Available tools: portal, search, FAQ, chat, … Dependability and Access Control Low cost of usage

8. University of Southern California Center for Software Engineering C S E USC August 2001©USC-CSE8 Proposed Metadata Categories Type of application (MIS, e-commerce, transportation, manufacturing, telecom, computer products, engineering & scientific, aerospace, public services, …) Application profile (see attached method pattern categories Developer skills and experience (students: undergraduate, grad, professional, well below average … well above average) Platform (mainframe, client, server, network, applications generator, …) Process Drivers (ranked): Requirements, cost, schedule, dependability, option exploration Size of project (maximum team size) Time Period (beginning, end dates; associated milestones) Experimental/Observational Assessment of Data Quality (weakly defined and reported … very thoroughly defined and reported)