90% of PDFs Now Easily Convertible to Text for Advanced Data Analysis

The process of converting PDFs to text has become significantly easier, with over 90% of PDFs now being convertible. This advancement has facilitated the creation of GraphRAGs, a type of Retrieval-Augmented Generation (RAG) system that uses graph data stores instead of traditional vector stores. GraphRAGs enhance data retrieval by incorporating reasoning capabilities. For instance, while vector store-backed RAGs can retrieve information like the name of XYZ, Inc.’s CFO from last year’s annual report, GraphRAGs can answer more complex queries. They can identify connections like which two directors of XYZ, Inc. attended the same school, without needing the school’s name in the query. This is possible because GraphRAGs construct a graph for data retrieval, a process detailed in a recent separate post.

Source: towardsdatascience.com