Pretraining and Finetuning a Transformer Model for Location Resolution

Discover how to extract and link location entities from unstructured text data in this 20-minute conference talk by Evelyn Wang, Data Scientist at Balyasny Asset Management. Learn about leveraging Databricks and the Hugging Face Transformers library to pretrain and finetune a custom model tailored for location resolution. Explore the process of creating training datasets, deployment considerations, and insights for processing locations in text data at scale. Gain valuable knowledge on adapting entity linking methods to domain-specific knowledge bases, particularly for fine-grained location resolution. This talk offers practical insights for those interested in enhancing location intelligence, risk management, and building knowledge graphs.