Word embeddings currently are one of the preferred representation for words in various NLP and IR tasks. In this talk, we will review the main embeddings used in IR and see to which extent they lead to improved IR performance. We will also discuss the possibility to extend current word embeddings with syntactic information and see the impact of doing so on several NLP tasks. Prof. Eric Gaussier is known for his work on the intersection of Arti cial Intelligence (AI) and Data Science (DS), in particular for his contributions on models and alogrithms to extract information, insights and knowledge from data in various forms. He has worked on three main sub elds of AI and DS: machine learning, information retrieval and computational linguistics. He is also interested in modeling how (textual) information is shared in social (content) networks, and how such networks evolve over time. More recently, He has also been working on improving job scheduling techniques through machine learning, and in learning representations for di erent types of sequences, as texts and Time series.