pdf - How to automate testing where the quality of results is largely subjective -
I am developing an automation and statistical generation program that is a fast prototype to see through a series Is that the licensing cost of the API is generated? A good return on investment is not a limited time limit and no useful tool with my limited experience, which has no automatic quality or accuracy testing and matrix.
This program works with the localization process for PDF documents. Specifically, it produces a report on some content of the file (approximate word count, image count, etc.) and has some content extraction and processing features. It is mainly used to reduce the time and cost to evaluate the cost of PDF localization project.
The application has now been approved for a more formal development process, in which the bug tracking system initial test, release and response cycle.
Then the question is how can you go about QA and testing in such an application, where numbers are not always useful due to the terrible construction of some estimated and processed output source documents? I am planning to add reports' warnings when the numbers clearly become insane, but what can be done to guarantee quality?
The most sophisticated solution so far has been to guarantee the results of some helpful methods, claim testing in the build environment and writing a bunch of traditional user trial cases (which I like to save).
How do you test for subjective quality remedies?
I am working in C #, but I support a common best practice which gives a response on any type of structure.
I'm not sure what application you are making, but to answer the common question: Create a collection of test cases that reflect the exchanges of your borders and see if it can properly judge those people. You can not actually test the actual input documents.
And then, there is a point where you have to accept what can be accomplished with automatic testing. When things really become subjective aesthetics or usable, for example) you will need to make a suitable decision for a real human.
I think I can give more useful answers.
Comments
Post a Comment