As part of Microsoft’s AI at Scale effort, we are leaning heavily on recent developments in large Transformer-based language models to improve the relevance quality of search results. This allows search engines to go beyond keyword matching to searching using the semantic meaning behind words and content. At Ignite (March 2021) we announced that Microsoft is bringing semantic search models developed by Microsoft Bing into Azure Cognitive Search. 

This talk will discuss the details of why transformer models are so effective at NLP tasks and give a deep dive into how we built and transferred the models to Azure.

Instructor's Bio

Alec Berntson

Principal Engineering Manager at Microsoft

Alec is a Principal Engineering Manager on the Search and Artificial Intelligence team at Microsoft. His team’s mission is to help other Microsoft products incorporate cutting-edge search technology such as semantic ranking, machine reading comprehension, extractive summarization, and classification filters. The team puts in a lot of work to make AI research ‘industrial strength’ for cost-effective and performant deployment across the world; then brings AI to scale.

Local ODSC chapter in Dublin, Ireland


  • 1

    Taming NLP Model Zoo: Bringing power of transformer models in Bing to semantic search in Azure Cognitive Search

    • AI+ Training

    • Webinar recording

    • AI+ Subscription Plans