Projects

RENFO

International expert mining of web search results with deep reinforcement learning

We are currently using reinforcement learning on biographical web search results to track professional mobility of international experts. We suppose that mining the web with semantic methods could provide valuable data for migration sociologist studying highly qualified migrations, and furthermore it could allow the development of a new generation of expert mining systems based on artificial intelligence.

Funding: Labex EFL (2017-2023) and Ecos Nord
Budget: ~43k€
Consortium: LIPN, Télécom-Paristech, IRD, IIMAS/UNAM, CRIM/UNAM

#GenMicFic

Artificial Intelligence for flash fiction generation

This project aims at generating flash fiction of fewer than 300 words with Transformers and Reinforcement Learning models. We test different methods for fine-tuning existing pre-trained language models with a flash fiction corpus in Spanish, French, and English. We evaluate the productions of these models both by standard readers and literature scholars. The goal of the project is to improve automated generated flash fictions and to empower human creativity with artificial intelligence tools.

Funding: Ecos Nord (2022-2025)
Budget: 60k€
Consortium: LIPN, Tec de Monterrey, IIMAS/UNAM (Mexico)

Publications

GeSERA: General-domain Summary Evaluation by Relevance Analysis

López Espejel, J., de Chalendar, G., Garcia Flores, J.J. (September, 2021) Proceedings of the International Conference on Recent Advances in Natural Language Processing (RANLP 2021), 856-867.

Saucissonnage of Long Sequences into a Multi-encoder for Neural Text Summarization with Transformers

López Espejel, J., de Chalendar, G., Garcia Flores, J., Meza Ruiz, I, and Charnois, T. (January, 2021) Extraction et Gestion des Connaissances (EGC), Montpellier, France,, Jan 2021, Montpellier, France.

Apprentissage par renforcement pour la recherche d’experts sur le web

Alizadeh, P. Garcia Flores, J., Meza Ruiz, I. (January, 2020) Extraction et Gestion des Connaissances (EGC’2020), Brussels, Belgium.

Recommendations on automatic document analysis : acquisition, management, exploration

Nedellec, C., Nazarenko, A., [...] Flores, J. [...] and Zweigenbaum, P. (September, 2019) Rapport de recherche : Comité pour la science ouverte , 12p, Paris.

Towards Identifying for Evidence of Brain Drain from Web Search Results using Reinforcement Learning

Murrieta, H., Meza, I., Alizadeh, P. and Garcia Flores, J. (December, 2019) LatinX in AI Research Workshop at the 33rd Conference on Neural Information Processing Systems (NeurIPS 2019) Vancouver, Canada

Robot Experience Stories: first person generation of robotic task narratives in SitLog

Garcia Flores, J., Meza, I., Colin, E., Gardent, C., Gangemi, A. and Pineda, L. (May, 2018) Journal of Intelligent and Fuzzy Systems. 34(5), pp.3291-3300, IOS Press

more

Code

Name Description Role URL source URL prototype
UnoporunO Web mining of highly qualified migrations project manager, developer https://github.com/rcln/unoporuno http://tal.lipn.univ-paris13.fr/unoporuno/
unoporunoDQN Reinforcement learning for highly qualified migrations tracking co-project manager https://github.com/rcln/unoporunoDQN
cartographies sonores World languages audio map of some of the languages studied by Labex EFL researchers. project manager https://github.com/rcln/unoporunoDQN http://tal.lipn.univ-paris13.fr/cartographies/
BNI Visual navigation on selected philosopher's ideas co-project manager https://github.com/rcln/bni http://tal.lipn.univ-paris13.fr/bni/
CCTV Multilingual Wikipedia topic visualization app for a sampled min-hashing topic extraction method. project manager https://github.com/rcln/min-hashing http://tal.lipn.univ-paris13.fr/minhashing/
Golfred Robot experience stories generation project manager, co-developer https://github.com/rcln/golfred
SOPA-Semeval Linear regression involving a soup of features for calculating semantic similarity between a pair of sentences (SEMEVAL-STS 2013-2015 system). co-developer https://github.com/rcln/semeval

Students & teaching

Université Paris 13, Institut Galilée
Master in Computer Science EID2

Project Management

Theoretical and practical course on project management for Master-1 students of Galilée Engineering School, at Sorbonne Paris Nord University.

Jessica López Espejel (LIPN, USPN)

I was Jessica's co-supervisor on her PhD (2019-2021) on Automatic summarization for long medical documents using multi-encoder Transformers. She also worked on semantic evaluation metrics for automatic summarization based on Wikipedia.

Josué Urbina and Carl Posthuma (Engineering College, IIMAS, UNAM)

I was Josué's and Carl's co-supervisor of their professional practices at UNAM Engineering College. They worked on a Deep-Q neural network based method for extracting biographical data from web search queries for expert mining.