Paper

https://github.com/keypointt/reading/blob/master/facebookResearch/crowd-intelligence-enhances-automated-mobile-testing.pdf

note

Crowdsourced Mobile Testing - using nonprofessional crowd testers

POLARIZ, proposed in this paper, which collects and analyses test inputs from a crowd call to non-technical users with no specific software testing expertise or experience.

  • evidence that motifs extracted from crowdsourced tests can complement and extend state-of-the-art automated test generation (via SAPIENZ)

SAPIENZ, baseline comparator, https://github.com/Rhapsod/sapienz

workflow, basically connecting crowd and apps-to-be-tested

workflow

workflow details

  1. App developer/researcher who seeks mobile test automation enhancement;
  2. Crowd workers/testers;
  3. The intermediary platform (i.e., POLARIZ platform) on which the crowd works.

Crowd Motif Extraction Algorithm

algorithm

  1. this algorithm is like a PCA or something, get most important motifs as testing pattern and remove duplciated ones.
  2. inspired by DNA sequence/motif searching.

Reference: