Alexandre Salle

Staff Data Scientist at Shopify; Ph.D. in NLP

alex@alexsalle.com

LinkedIn

GitHub

Education

Universidade Federal do Rio Grande do Sul — Porto Alegre, Brazil
2015-2021
Ph.D. - Natural Language Processing - Distributional Semantic Models

ENSIMAG - Grenoble INP — Grenoble, France
2012-2014
Double-degree program with the Universidade Federal do Rio Grande do Sul, receiving the French “Diplôme d’ingénieur”.
Studied Information Systems Engineering and Parallel, Distributed, and Embedded Systems.

Universidade Federal do Rio Grande do Sul — Porto Alegre, Brazil
2010-2014
BSc. Computer Engineering

Publications

ChatGPT Goes Shopping: LLMs Can Predict Relevance in eCommerce Search (pdf)
Beatriz Soviero, Daniel Kuhn, Alexandre Salle, and Viviane P. Moreira
ECIR (2024)

Native Language Identification with Large Language Models (pdf)
Wei Zhang and Alexandre Salle
arXiv (2023)

A Simple Loss Function for Convergent Algorithm Synthesis using RNNs (pdf)
Alexandre Salle and Shervin Malmasi
ICLR (2023)

Understanding the effects of negative (and positive) pointwise mutual information on word vectors (pdf)
Alexandre Salle and Aline Villavicencio
Journal of Experimental & Theoretical Artificial Intelligence (2022)

CoSearcher: Studying the Effectiveness of Conversational Search Refinement and Clarification through User Simulation (pdf)
Alexandre Salle, Shervin Malmasi, Oleg Rokhlenko, and Eugene Agichtein
Information Retrieval Journal (2022)

The role of negative information when learning dense word vectors (pdf)
Alexandre Salle
Ph.D. Thesis (2021)

Studying the Effectiveness of Conversational Search Refinement Through User Simulation (pdf)
Alexandre Salle, Shervin Malmasi, Oleg Rokhlenko, and Eugene Agichtein
ECIR (2021) - Won Industry Impact Award

Why So Down? The Role of Negative (and Positive) Pointwise Mutual Information in Distributional Semantics (pdf)
Alexandre Salle and Aline Villavicencio
arXiv (2019)

Think Again Networks and the Delta Loss (pdf)
Alexandre Salle and Marcelo Prates
arXiv (2019)

Restricted Recurrent Neural Tensor Networks: Exploiting Word Frequency and Compositionality (pdf)
Alexandre Salle and Aline Villavicencio
ACL (2018)

Incorporating Subword Information into Matrix Factorization Word Embeddings (pdf)
Alexandre Salle and Aline Villavicencio
SCLeM at NAACL (2018)

Enhancing the LexVec Distributed Word Representation Model Using Positional Contexts and External Memory (pdf)
Alexandre Salle, Marco Idiart, and Aline Villavicencio
arXiv (2016)

Matrix Factorization using Window Sampling and Negative Sampling for Improved Word Representations (pdf)
Alexandre Salle, Marco Idiart, and Aline Villavicencio
ACL (2016)

Recommendation-assisted Playlist Creation (pdf)
Alexandre Salle
BSc. thesis (2014)

Software and Resources

 

Random Projects (from newest to very, very old!)