This spring semester, I am working on a data mining research relating to software testing and parallel computing. Some of my current research questions:
- how can data mining help software testing?
- what are the challenge of software testing? as of now, I know that the input space for testing - or test space is infinitive - thus testing will be complicated and expensive - where I think data mining tech come into help by reducing this infinite search space into a cost effective critical small subset of tests needed?
- what is the most minimal test set- test requirement? i think a lot of research and study has come up with quite some metrics of test requirement, coverage for this... the next question is how to evaluate goodness of 1 test set to others? = how to evaluate goodness of 1 cluster of tests with another cluster? by what metrics?
- as of now, I envision:
testing challenge = infinite state input space, infinite and costly tests
desired outcome= a group of minimalist cost effective critical tests
tool = data mining clustering algorithm
project scope: clustering analysis of test space
s1= for imperative program
s2=logic programming/finite state machine
s3 = parallel and concurrent program
most of the reference library books I have read so far all talking about logics, concurrent programming and testing. The question that bothering me is what is the relation between logic and testing and concurrency?????????????????? why what what do I have to deal with logic???
