Skip to main content

User login

What is not?

 Impossible is not a fact, its an opinion 

Milestone map

May 23, 2010 by dangyogi

I would like to record the steps and capabilities needed so that we can start sorting them into a plan.

Here is a starting list.  Please add/change/delete to add your own ideas or corrections.  If the site will let you, feel free to edit this post directly, other post a comment and I will try to perodically incorporate the comments into this list.  (Note, this list is not sorted yet).

  • CLI interface
  • Web interface
  • GUI interface
  • Dialog for entering initial presenting symptoms.
  • Dialog for additional symptoms.
  • Dialog for exam.
  • Dialog for investigations.
  • Dialog for treatment.
  • Alert mechanism.
  • CF mechanisms.
  • Web interface for entering/editing rules.
  • Automated tools to extract text for translation
  • Integration with standard translation tools
  • Automated tools to convert the rules entered through the web interface into .krb files.
  • Figure out an automated test harness for the rules
  • Figure out a test harness for the software

Here's a first shot at a milestones list:

Milestone 1: Figuring out CF mechanism is primary goal

  • CF mechanism
  • Automated test harness for rules (means of achieving that goal)
  • May want tools to produce K-Maps to confer w/experts.  If so, these should also serve as the automated test inputs.
  • Perhaps no UI (not even CLI), just use test harness to run the rules and report the results
  • Probably only need ~10 diseases to start with in one .krb file
  • Don't need phases of questioning: symptoms, exam, investigations.
  • Don't need text for questions, just use internal q_names.
  • Don't need translation

Milestone 2: Figuring out primary user interaction w/questions is primary goal

  • Primary user interaction w/questions
  • CLI UI
  • Set this mechanism up to handle language translation
  • Have questions in two languages (different alphabets) for testing.  More translations not need yet.
  • Still only one phase of consultation, maybe the "exam" phase?
  • Still only ~20 diseases

Milestone 3: Add phases of consultation.

  • Figure out how to add phases of consultation
  • Figure out how to construct initial "canned" question list after entering presenting symptoms
  • Still CLI
  • Still only ~20 diseases, but adding the rules for each phase now
  • Figure out alert mechanism

Milestone 4: Web UI

  • Build a very basic Web UI and make it available to everyone.
  • SF might not be the best place for this.  They seem limited to a CGI interface.  So we may need to get our own server set up somewhere.
  • We should be working on a test harness for the software at this point too.
  • Figure out how to scale the number of rules into the thousands.  What does each .krb file represent, and what directory structure do we use.
  • This is probably the point that we need to ask ourselves if we really still need Pyke.  Can this be done just using a relational database?

Milestone 5: Mechanism to gather rules

  • Figure out how to gather rules on the internet
  • Figure out how to automatically translate these rules into .krb files
  • Figure out how to include test suites in the rule gathering
  • Figure out how to automate the generation of .pot files for translation
  • Figure out how to run the test suites and review the results through the web interface
  • Figure out how we are going to version rules
  • Figure out how we are going to promote rules into the production system

Milestone 6: Deployment

  • Figure out how to deploy the system to the world
  • Develop a GUI interface (if that's still seen as what we need)
Premium Drupal Themes by Adaptivethemes