Hi, I'm Mathilde. I come from Alsace, a beautiful part of France where I did my two first years of Bachelor at the University of Strasbourg. After a research internship, I questioned my academic background and decided to leave France for a year to improve my english and think twice about my objectives. I was an aupair for 8 months and really improved my english which was a pain for me1. Thanks to my host-family.
I may look brave for leaving university and having a step back but I was not so much. Indeed, I continued my Bachelor by distance learning with the University of Aix-Marseille. As I had to split the academic courses in two years, I did an internship during 6 months in Strasbourg during the second year.
Then I decided to move in Lyon for the specialities of the Master degree. I choose "Data Science" and finally discovered what really interests me.
Since November 2018 I am a PhD student at - LSV, ENS Paris-Saclay, Inria - under the supervision of T.Chatain and J.Carmona. My research focuses on conformance checking and model repair in Process Mining.
 I have hearing problems and that does not help for learning languages. My apologises for asking to repeat.
Process Mining is a growing field that sits at the intersection between data and process science. Techniques in this field are grounded on using the evidences on the process executions to mine, analyze or improve process models. Being a young field, there exist several problems that have not been considered so far, representing a real challenge for current process mining algorithms. A satisfactory addressing of some of these problems will clearly widen the applicability of process mining techniques in practice. During my PhD, I focus my work in considering some of these challenges. The first challenge to consider is the alignment between reality and process models, i.e., to obtain model' explanations of the reality, as it has been recorded in an event log. This is known to be a complex problem, due to the well-known state explosion problem. Furthermore, the incorporation of other perspectives like timestamps and data attributes makes the problem even more challenging. During this PhD, the development of disruptive techniques to compute alignments in a general setting will be proposed. The second challenge is optimizing the repair of process models. Here the motivation is to use the deviations detected in alignments to improve the process. Current techniques for this task only propose local, peephole, optimizations, that lack a global guarantee. The thesis will explore fresh ways of repairing process models so that represent an optimal solution to the problem.