MetaGraph: knowledge graphs from financial NLP

This September 2025 paper presents MetaGraph, a novel methodology for constructing knowledge graphs from scientific literature, specifically applied to Financial Natural Language Processing (NLP) research between 2022 and 2025. The authors utilized Large Language Models (LLMs) to extract key information from 681 papers, including tasks, datasets, models, motivations, and limitations, and organized it into a structured, queryable format. The analysis highlights three phases in Financial NLP's evolution: initial LLM adoption and task/dataset innovation, subsequent critical reflection on LLM limitations, and a current trend toward integrating peripheral techniques into modular systems. The research reveals a shift toward Financial Question Answering (QA), increased use of synthetic data, and a growing emphasis on open-source models and system-level solutions like Retrieval-Augmented Generation (RAG). Ultimately, MetaGraph offers a reusable framework for quantitatively mapping scientific progress and understanding changing priorities in a rapidly evolving field.Source:https://www.arxiv.org/pdf/2509.09544

There are no comments yet.

Please log in to write the first comment.