Launched at the beginning of the 21rst century, PISA surveys (Programme for international student assessment) whose have been a part of a specific European context, whose purpose is to establish a European policy in the education field, and are characterized by the development of a soft governance through benchmarking, that is a supposed suggestive one only, based on comparative indicators, supposed to foster a sound emulation. After having rapidly described this context, the paper will recall the main technical characteristics of these surveys, before developing the following discussion: can they claim to make up perfect tools for international comparisons of educational systems, and/or an efficient lever for some changes or policies concerning these systems?

Introduced in the 1980s, the concept of content pedagogical knowledge (PCK) has grown significantly in the field of mathematics in the next decades. From the outset, PCK assessment has attracted the attention of researchers. After reminding the origin of the concept, two assessment approaches (static vs dynamic) of PCK are put into perspective. Given the feature of preschool, PCK assessment of teachers practicing at this level is developed in the second part of the paper. Finally we will see that the question of PCK is intrinsically linked to that of teachers' professional development. In this domain, recent research opens new and interesting perspectives, especially in the field of assessment.

This article presents three important dimensions to consider when analyzing the professional integration performance of professional graduates: the diversity and professionalization nature of young people's pathways upstream of their diploma, the selectivity of professional education courses and the types of labor market covered by these degrees. Although not related to the quality of the education provided, these three dimensions can explain much of the "good results" of a professional degree. This leads us to be very cautious about the evaluation of a professional degree through its insertion performance.

For thirty years, natural language processing has given more and more importance to machine learning, resulting in technological progress for an ever increasing number of applications. These advances have been made possible by the availability of corpora as learning material, the development of the evaluation paradigm (shared tasks) and the creation of infrastructures for technology evaluation. But copora need to be of appropriate size to be representative of the linguistic reality and neither the needed resources nor the means to produce them exist for all languages. Experiments have shown that the size of the learning material can be a substitute for algorithmic complexity or expert knowledge, but have we reached the limit for progress with this kind of approach?