This tool uses LangChain’s Pydantic parser to identify crucial entities from an input user query to automate the process of dataset filtering, based on a pre-defined data schema. With this automated unstructured to structured data manipulation technique, we can extend this tool to various internal initiatives such as SoW categorization, Resume parser and even client-facing problem statements such as marketing audience filtration.
Within this code and demo, we have outlined two use cases:
- Audience Filtration: Here, our use case is that we can perform Text-to-SQL functionality on any given audience data with the help of this parser
- Product Categorization: Here, our use case is to perform Text-to-JSON functionality on a given digital media content dataset using this parser. This has also been implemented for NBCU client as a standalone app, we hope to present this in their upcoming Gen AI workshop in November.
Screenshot Preview