%%%%%%%%%%%%%%%%%%%%% POSTDOCTORAL POSITION LIFO, Université d'Orléans, France %%%%%%%%%%%%%%%%%%%%% + Title: Data Science Queries: Language and Algorithms + Founding: Regional project DOING (APR-IA, Region Centre Val de Loire) + Starting: 01/10/2022 (approximative) + Length: 12 months. + Research laboratory: LIFO (https://www.univ-orleans.fr/lifo/?lang=en) (PAMDA team), Université d'Orléans, France + Collaboration with the following laboratories: ---- LIFAT (https://lifat.univ-tours.fr/) and ---- LIRIS (https://liris.cnrs.fr/ + Salarie : 2000€ net. %%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%% CONTEXT %%%%%%%%%%%%%%%%%%%%% THE REGIONAL PROJECT: DOING project aims to develop methods and tools to first extract information from textual data by structuring it in a database graph, and then to manipulate this knowledge graph in an intelligent way. The chosen application domain is the health domain, primarily using freely available data (such as clinical cases). DOING aims to design a new form of declarative queries, which can integrate analyses, that will guide health specialists in their decision-making. DOING (https://www.univ-orleans.fr/lifo/evenements/doing/) is based on a real interdisciplinary collaboration to transform data into information and then into knowledge. This is a project in the context of DOING@MADICS action and DOING@DIAMS working group. %%%%%%%%%%%%%%%%%%%%% SUBJECT %%%%%%%%%%%%%%%%%%%%% The postdoctoral research work concerns Task 2 of the DOING (APR-IA) project. The goal is the development of a first version of a query system on graph databases, whose (declarative) query language would encompass predictive analysis - a term that combines data management, machine learning and optimisation, and which reflects the growing demand for such tools for handling data science problems. The task consists of the following steps: (1) Modelling a query language on a graph database for the application of a graph node classification method. There are many applications in the health domain where the question is to predict the labels of the nodes of a graph (starting with only a fraction of the labelled nodes). For example, predicting whether a patient should be screened for cancer, based on his or her health record, and the similarity of his or her case to other patients. Solutions for node classification using unsupervised methods or convolutional networks exist. The aim is to evaluate their use in the context of databases where only a portion of the graph is available at a time. In particular, we wish to study the problem of prediction by node classification via deep learning techniques. We wish to propose an adaptation of these methods for integration into the Cypher language (Neo4J DBMS). (2) Method for the construction of query execution plans and study of possible optimisations. The aim here is to prepare the data science pipeline necessary for the preparation of data and the application of learning methods by integrating constraints. %%%%%%%%%%%%%%%%% COLLABORATION TEAM %%%%%%%%%%%%%%%%% The postdoctoral proposal is done in the context of a collaborative work which involves the following researchers: At LIFO: (1) Jacques Chabin (2) Mirian Halfeld Ferrari At LIFAT (1) Donatello Conte (2) Jean-Yves Ramel At LIRIS Genoveva Vargas Solar The postdoctoral fellow will have the opportunity to interact to all the group. %%%%%%%%%%%%%%%%% CANDIDATE PROFILE %%%%%%%%%%%%%%%%% The candidate should have a PhD degree in computer science. The work involves knowledge in the domain of databases and of machine learning. Skills on at least ONE of these areas are required. The candidate should also be motivated to invest in the complementary field. A good English level is also required. French is not mandatory for candidates with a very good level of English and willing to learn French for daily life in France. The research work is conducted at the Laboratoire d'Informatique Fondamentale d'Orléans (LIFO), in France. The postdoctoral fellow should be physically present (i.e., the Postdoctoral position is not achievable by remote work). %%%%%%%%%%%%%%%%% TO APPLY %%%%%%%%%%%%%%%%% Send your applications by email to the following addresses: (1) mirian@univ-orleans.fr (2) donatello.conte@univ-tours.fr The application should contain: + a detailed curriculum vitae (with your email), + the Phd diploma + reports of pre-defense (if you have a French thesis) + MSc transcripts + two references We set up selection cycles (with interview periods) as the first step. We plan the first round of interviews at JULY (applications needed before 10 JULY).