MICoSys Speaker Series (2023-2024)

Dr. Leonardo Lezcano, Sr. Data Science Manager at Walmart Global Tech

Thursday, October 26 at 3:00 pm

ISB 130 

Interdisciplinary Science Building @ SJSU

Cross-lingual Search for e-Commerce based on Query Translatability and Mixed-Domain Fine-Tuning

Short Bio:

Dr. Leo Lezcano is a Sr. Data Science Manager at Walmart Global Tech where he leads the Spanish Experience initiative and other multilingual efforts. Originally from Cuba, he holds a PhD in Computer Science from the University of Alcalá, Spain, followed by a post-doc at the Open University of the Netherlands. Leo moved to Bay Area in 2016 attracted by Big Data and Machine Learning challenges. He introduced the first Apache Spark production pipeline at eBay Search Science. Then joined VUDU to develop its movie recommendation system. Leo's experience on information retrieval expanded when he joined the Query Understanding team at Walmart Labs. Managing a small team of NMT experts, he envisioned and launched the first multilingual query support at Walmart Search. His publications include the textbook on Language Processing for the UDIMA University in Madrid, and more than 30 research papers in the domain. Leo is in constant pursue of new data driven experiences that improve lives and business.

Talk Abstract:

Online stores in the US offer a unique scenario for Cross-Lingual Information Retrieval due to the frequent mix of Spanish and English in queries. While physical stores have traditionally allowed a visual product search by walking the aisles, the hyper e-Commerce adoption driven by the COVID pandemic has turned the lack of familiarity with specific English terms into an obstacle for a smooth online experience. Machine Translation can lift query relevance in these scenarios, but context scarcity, language ambiguity, polysemy-derived problems and other domain specific obstacles, make generic MT an impractical solution. In this talk, we will be covering the multilingual challenges that e-Commerce websites deal with in their domain. For example, beyond raw MT accuracy, and given that it is not the final goal but the means to a frictionless buying experience, the problem of query translatability will be explored. Overall, it will be a good opportunity to go over the everyday life in companies with a big tech ecosystem, what's happening inside and outside your laptop, as well as skills that are foundational to navigate the Big Data and AI era.