====== Notes from 09.09.2010 ======

===== Brainstorming Notes =====

  * Data values
    * From DB
    * Machine learning for equivalence classes
    * Separate models for valid or invalid values
  * Can we map navigation model to code?  How precise?
    * Did we not cover edges because of probabilities or because not representative?
  * Navigation Model: Ways to create
    * by user
    * by time
    * by type of user
    * group by purpose/intention --> how?
    * length of user sessions?
  * How to define "critical" pages?
  * What makes a test case "feasible", or, alternatively not feasible?
    * cross-accessible
    * what kind of distribution of test cases do want?  (What % feasible, what % not?)
    * not feasible: x # of error responses in a row --> not feasible
  * Can we define "critical pages"?
    * non-error response?
    * how many pages are critical? (i.e., aren't error responses)
  * What are properties of the user sessions we may want to look at?
    * if it changes the state
    * length of user session
    * timing --> take longer on a page
    * # of parameters?
    * previous page, # of parameters --> values change
    * the user's intent (browsing, vs something else) --> define equivalence classes

~~DISCUSSION~~