EN FR
EN FR


Section: New Results

Crowdmining to Increase the Quality of Software Systems

Modern software systems, especially in the open source world, are more and more part of ecosystems where large quantities of data about these systems are available. These data may come for example from application stores (e.g. Google Play Store or Apple Store for mobile applications), forges (e.g. GitHub), or from the usage conditions experienced by users of these software systems. This large amount of data enables to unlock some specific challenges where knowledge about the software systems can be automatically mined and learnt. In this domain, we obtained new results on the mining of mobile software antipatterns on a crowd of mobile applications and their versions to study their impact on resource consumption [32]. This result has been achieved in the context of the PhD thesis, defended in November 2016, of Geoffrey Hecht [13]. We also consider the crowd of mobile devices and users to detect and reproduce application crashes in the wild. By leveraging our results in the domain of in-breath monitoring, we use the APISENSE ® platform (see Section 6.1) to collect extended crash reports that can be aggregated to infer the minimal execution path that lead to a crash [28]. This result has been achieved in the context of the PhD thesis, defended in December 2016, of María Gomez Lacruz [12]. These results are also in relation with our activities in the context of the SOMCA associated team (see Section 9.4).