*Laboratoire d'Informatique de Paris Nord <https://lipn.univ-paris13.fr/en/home/> / Université Sorbonne Paris Nord* *Closing Date: April 15 2020*
The RENFO project (Deep Reinforcement Learning for Expert Tracking from Web Search Results) funded by the Empirical Foundations of Linguistics (EFL) laboratory of excellence and the ECOS Nord fund is looking for a PhD candidate to develop deep reinforcement learning methods for web search query results in order to find experts from a certain scientific discipline, a country of origin or a person’s name by mining the web.
Most of current expert finding approaches work on highly organized data sources, like scientific publications or the semantic web. We propose an original approach based on information extraction from a noisier data source: web search query results. Deep Reinforcement Learning (DRL) has shown promising results on information extraction from web search results. However, the association of web search results to experts, and the extraction of meaningful semantic information (affiliation institution, city, year, scientific discipline) from the resulting dataset is challenging both from a scientific and a technological perspective. As web search querying is expensive from an algorithmic point of view, reinforcement learning is intended to optimize the query formulation process and result exploration policy with a Deep-Q Network.
The goal of the PhD will be to provide an end to end DRL model for expert finding, while integrating state of the art information extraction strategies to the method. This approach will be evaluated on an existing dataset of international experts coming from developing countries and having obtained a scientific degree abroad. An ongoing expert finding system is currently being development, and the PhD candidate is expected to participate in this development.
While most of the PhD will be carried out at the Northern Paris Computer Science Lab (LIPN) at Sorbonne Paris Nord University (Villetaneuse campus), the PhD student will perform yearly scientific of one or two months at UNAM, funded by an an ongoing ECOS Nord project in association with Mexico.Skills required
→ Sound background in computer science → Familiar with deep learning and information extraction and Natural Language Processing → Excellent programming skills (Python, and some Pythorch or Tensorflow experience) → Autonomy, creativity, curiosity → Master’s degree in Computer Science or Natural Language Processing → Full proficiency (spoken, written) in English → Any French and Spanish skills would be appreciated
Send motivational cover letter and CV before April 15th to:
Jorge Garcia Flores / jgflores at lipn.univ-paris13.fr Pegah Alizadeh / pegah.alizadeh at devinci.fr Iván Vladimir Meza / ivanvladimir at turing.iimas.unam.mx -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/html Size: 3304 bytes Desc: not available URL: <https://mailman.uib.no/public/corpora/attachments/20200214/605c6fa2/attachment.txt>