Natural Languages Processing and Knowledge Extraction


The NLP team at ATLC is focusing on building unsupervised pipelines for Text processing targeting a broad spectrum of tasks ranging from Input method editors to large-scale knowledge extraction.

The team started addressing Arabic-specific problems and has built a comprehensive Arabic NLP Stack that was a pillar to improve key features across different products, in particular in Search, both Enterprise and Web, Translation, Proofing Tools, IMEs and Speech. Then more recently the team focus shifted to address Knowledge extraction problems aiming at building full pipelines that are scalable on both language and domain dimensions, and at the same time unsupervised or weakly-supervised.

Many of the team’s projects have shipped into products including Office Word, SharePoint, Bing, PowerPoint, Satori, Exchange, Windows Phone…