Jingwen Lu
|
Principal Applied Science Manager
Microsoft |
|
I’m Principal Applied Science Manager in Search and Artificial Intelligence at Microsoft. I am a Applied Science leader who drives key areas of “query rewriting” for Microsoft Bing. My team is responsible for various systems that search users take for granted, ranging from spell correction, synonym expansion, and query intent classification and rewriting, that help you discover rich content on the web. Microsoft has deeply invested in the deep learning wave, with in-house, very large scale pre-trained models for both NLP and natural language generation. We apply state-of-the-art models in production to bring better search results to Microsoft users
|
Workshop: Spelling Correction for 100+ Languages
Prerequisites: Assuming some familiarity with: Edit distance & Language model
Other concepts that will be used: Lexicon, Noisy channel, Deep learning, Zero-shot learning, Language family
We likely have all found ourselves to spell words wrong when we compose essays and documents, write emails and text messages, conduct searches on search engines and etc. We have also been accustomed to have an automatic spelling correction service that helps with these tasks. Spelling correction has been one of the classic Natural language Process problems for quite sometime. In this workshop, we will take a detailed tour in how various algorithms spelling correction models are built on and how latest advances in Machine Learning, especially Deep Learning, have greatly helped us to build quality services for 100 plus languages.
Other concepts that will be used: Lexicon, Noisy channel, Deep learning, Zero-shot learning, Language family
- Speller100: Zero-shot spelling correction at scale for 100-plus languages - Microsoft Research at this link: https://www.microsoft.com/en-us/research/blog/speller100-zero-shot-spelling-correction-at-scale-for-100-plus-languages/
We likely have all found ourselves to spell words wrong when we compose essays and documents, write emails and text messages, conduct searches on search engines and etc. We have also been accustomed to have an automatic spelling correction service that helps with these tasks. Spelling correction has been one of the classic Natural language Process problems for quite sometime. In this workshop, we will take a detailed tour in how various algorithms spelling correction models are built on and how latest advances in Machine Learning, especially Deep Learning, have greatly helped us to build quality services for 100 plus languages.