PPTX 1273 KB

Report
Curriculum Vitae Management System
Taming the CV Data Beast
Robert (Bob) Dalgleish, Systems Analyst
Alan Deschner, Senior Analyst and Project Manager
Information & Communications Technology
Noises Off
Help reduce interruptions.
Please turn off the beeps and boops
from your portable devices.
The Project
Alan Deschner
Senior Analyst, ICT Applications
Project Manager, CVMS Project
Early History
• 2007, 2008 – Various colleges at U of S
investigated or partially built CV systems
• Oct’08 – College of Medicine:
– Formed eCV Task Force
– Driven by costs of accreditation reporting
– Wrote project charter for software development
• Fall’08 thru Spring’09
– Requirements gathering and analysis
Options and Decisions
• Considered commercial systems, other
Universities’ systems, other colleges at U of S
• Deficiencies (3 main ones)
– No support for U of S standard CV format
– Lack of import facilities (e.g., citations)
– Lack of integration with other in-house data
sources (student info, research info, etc.)
• Decision: develop in-house
History: Implementation Project
•
•
•
•
•
•
•
Feb’10 – proposal to develop in-house
April’10 – project start-up, analyst and P/A
July’10 – added another developer
Aug’10 – added ½ time project manager
May’11 – alpha testing
June’11 – beta testing
End of June’11 – ready to go-live, with a few
features not yet implemented
Production Go-Live
• June 28, 2011 – ready, but college decided
not to rush into production:
– Reduced availability in the college over the summer
– Desirability of additional features
• July and Aug’11 – extra development, testing,
remediation, and getting ready for go-live
• End of Aug’11 – live for College of Medicine
• End of Sept’11 – live for College of Nursing
Project Scope
• Release 1 (this project)
– Data entry screens
– Load data from EndNote, RefMan
– Produce U of S Comprehensive and Form 1 CV
– Interfaces to some internal systems (deferred)
• Releases 2 & 3 (the future)
– Reporting at department and college levels
– Publishing CV to funding agencies, websites,
document management system (workflow)
Project Organization
• Requirements gathering was fairly traditional
• Development used agile, iterative approach
• Use of Jira to manage the backlog, estimates,
iteration definitions
• Use of Confluence wiki for documentation
• eCV Task Force acted as the steering
committee
• Co-sponsors from Medicine and ICT
Agile Development
• Team was inexperienced in agile methods
• Daily 10-15 minute scrum
– Often took longer, sometimes ½ hour
– Hard to keep on track (inexperience)
• Started with 4-week iterations
– Started to vary, some stretched to 5 weeks, some
chopped to 3 weeks
– Got out of step with eCV Task Force meetings
User Involvement
• Representative from Medicine was assigned
to the project team.
– Administrative person who deals with CVs
– We wanted her as much as ½ time
– She was only available ½ - 1 day per week
• Many said, “You’re lucky to have that much!”
• Agile says, “You need a SME embedded in the
project team, full time if possible!”
Working Group
• A working group was put together from the
colleges and various departments
– Review and test the application deployed in the
previous iteration
– Identify new or changed functionality
– Help with details of the requirements
– Make recommendations to the eCV Task Force
• Met every second week, and did some
additional testing between meetings
Iteration Schedule
• In iteration 6, we started 2-week iterations
with defined reviews, releases to the test
environment.
Monday
Tuesday
Wednesday
Thursday
am: deploy N-1
pm: plan N
Testers review
iteration N-1
Last day to develop am: deploy N
iteration N
pm: plan N+1
Testers review
iteration N
Friday
Begin work on
iteration N
The Product
Bob Dalgleish
Systems Analyst, ICT Applications
Lead Analyst, CVMS Project
CV
CV
Administrative Staff
Faculty
Promotions and
Tenure Committee
Granting Agencies
Activity
Report
Research Services
Faculty
CVMS
Grad Studies
Citation
Management
Software
Course Information
Administrative Staff
Contact Database
Human Resources
Technical Architecture:
Application Stack
• RedHat Linux, JBoss EAP 5
CVMS
Application
Java
6 Virtual
Machine 3
Seam
2
Hibernate
Java 6 Virtual Machine
Java 6 Virtual Machine
And many more
Java 6 Virtual Machine
EAP
5
Java 6JBoss
Virtual
Machine
Java 6 Virtual Machine
RedHat Enterprise Linux
Virtual Machine Container
Deployment Process
•
Deployment: Workstations and Servers
Developer
Workstation
Subversion
Repository
Jira Issue
Repository
Jenkins
Continuous
Integration
TestDT
Test
Prod
Development Architecture
•
•
•
•
•
Eclipse (JBoss Developer Studio)
Subclipse (Subversion connector)
TestNG – test suite runner
Selenium – UI test suite developer & runner
Clover – test coverage reporter
– Goal was 70% coverage, stabilized at 63%
• Jira – Issue reporting tool, release management,
communications
• Confluence (wiki) – requirements & design capture
Technology Challenges
• Information Architecture:
– Information doesn’t mean the same thing to two
different people, departments or colleges
•
•
•
•
•
Standards, so many of them, so little time
Publishing: iText vs DocBook,
Hibernate as object repository
Categorization, for publishing and reports
Work flows
Information Architecture
• Many sources for data models and metadata
– CASRAI, Dublin Core, HR-XML.org, CommonCV.net
– Decided to normalize data and minimize structure clashes
• Working group process
– Discuss data elements, how they would be used, collected,
verified
– Built prototypes, tested by working group members
– Repeat
• Focused on requirements of university CV
– Publishing requirements
– Review process
IA – Imprecise Dates
IA – Field Lengths
Sample Published Output
Hibernate Data Handling
• Implements persistence (Java Persistence
Architecture)
– Push data to repository if it has been changed
– Coordinate network of linked objects
– Minimizes SQL to be written
• Inversion of Control
– LazyInitializationException - means you aren’t
where you think you are
– While Hibernate handles much of it …
Categories: Publishing …
• Each line item has a section selector
Categories: … Selecting
• Some items have many categories, some have
only one available.
Categories: … Published
• The section is where the item appears in our
standard CV …
Item Work Flow
• Each item goes through a sequence, Entered
(Not Approved), Requires Attention, Approved
Importing Citations
citations.xml
citations.txt
JAXB
Parser
EndNote
XML
Import
RefMan
Import
Import
Feedback
CVMS
Next Steps
Phase I is complete for two colleges, what now?
• Next biggest need is reporting
• Complete some of the deferred features
• Integrate with other systems
• Roll out to the rest of campus
This looks like an on-going program.
Next Steps - Reporting
• Reporting for departments and colleges
– Faculty activity reports
– Research reports
– Reports for accreditation
• User-defined query, extract, reporting facility
– Facilitate external analysis
Reporting
Reporting
Audiences
Faculty & Committees
Accreditation Panels
…
Data
Reporting
Reporting
Data
Governance
Data Quality
Management
Audiences
Processes
Faculty & Committees
Accreditation Panels
…
People
Reporting
Data Quality Management
- Incorrect data
- Missing data
- Redundant or duplicated data
Report Parameters
- Periods
- Categories
- Populations
Report Format
- Raw Excel file
- Formatted Word or HTML file
- Dashboard
Next Steps – CV Publishing
• Publish CV to document management system
– Enable workflows for college reviews related to
tenure and promotion
•
•
•
•
Publish to personal website
Publish to commonCV.net
Additional citation formats (only NLM now)
Additional CV formats
– Other funding agencies
Next Steps - Data
• Pull data from other U of S systems
– Student info, HR, research admin, grad studies
• Pull data from external systems
– e.g., PubMed, Scopus
• Bulk load of legacy data using cut-and-paste
and data parsing
• Additional data fields to support other
disciplines (e.g., humanities, fine arts)
Next Steps - Data
Lessons Learned
Lessons Learned
• Improve the change management processes:
– Changes identified by the working group need
approval from eCV Task Force, but there was not
enough time to answer the question,
“How much will it cost?”
– Scope creep is insidious! Beware!
– Help users understand that adding a new feature
means dropping something else.
Lessons Learned about Agile
• Include an expert agile practitioner when an
organizational goal is to change software
development methods to agile.
• Shorter iterations are better.
• The more user involvement, the better.
And constant involvement is best.
• Overall, agile is better than the alternatives!
Even More Lessons Learned
• Don’t expect users to know what they want
until they see what they don’t want!
• The great thing about standards is that there
are so many to choose from!
Questions?
Contact us:
[email protected]
[email protected]
http://www.usask.ca/its/services/cv_management_system/
Credits
Project Team
• Bob Dalgleish
• Boris Tchoursine
• Aileen Garcia
• Carla Zorn (Medicine)
• Alan Deschner (PM)
• Kirby McInnis (AIS Manager)
eCV Task Force & Work Group
• Dr Roger Pierson (co-chair)
• Dr Vivian Ramsden (co-chair)
• Dr Ron Doucette
• Dr Roger Keith
• Dr Angela Busch (PT)
• Dr Phil Woods (Nursing)
• John Costa
• Special guests

similar documents