[PhD defense] Embedding models for relational data analytics

2D embeddings of Wikipedia entities, colored by type.

The PhD defense of Alexis Cvetkov-Iliev on “Embedding models for relational data analytics” will take place on January 25 at 12:30 in Amphithéâtre Sophie Germain, Inria Saclay. It can also be attended remotely via the following link: https://inria.webex.com/meet/alexis.cvetkov-iliev.

Short abstract of the defense:
Data analysis, for instance with machine learning, typically requires data in a single table. In practice however, information is often scattered across multiple data sources that must be assembled together for analysis. To avoid this difficult and time-consuming step, we investigate here the potential of embedding models to facilitate the analysis of such unassembled data. We especially consider two problems: 1) data analysis across non-normalized sources, which typically requires entity matching (e.g. linking “Paris” and “Paris, FR”),  and 2) enriching data analyses with background information from external sources, which usually calls for tedious feature engineering.

Comments are closed.