Peter Baumann | Searcharxiv

arXiv2025

MetaGraph: A Large-Scale Meta-Analysis of GenAI in Financial NLP (2022-2025)

Financial NLP has evolved rapidly since late 2022, outpacing narrative surveys. We introduce MetaGraph, a methodology for extracting typed knowledge graphs from scientific corpora using ontology-guided LLM extraction to enable structured, large-scale trend analysis. Applied to 681 papers on GenAI in Finance (2022-2025), MetaGraph reveals three phases: early LLM-driven expansion of tasks and datasets, growing emphasis on limitations and risk, and a shift toward modular, system-oriented methods (e.g., retrieval-augmented designs). We release the resulting resource and artifacts to support reproducible meta-analysis and future monitoring of the field.

cs.CL↗

arXiv2024

Analyzing the coupling process of distributed mixed real-virtual prototypes

The ongoing connection and automation of vehicles leads to a closer interaction of the individual vehicle components, which demands for consideration throughout the entire development process. In the design phase, this is achieved through co-simulation of component models. However, complex co-simulation environments are rarely (re-)used in the verification and validation phases, in which mixed real-virtual prototypes (e.g. Hardware-in-the-Loop) are already available. One reason for this are coupling errors such as time-delays, which inevitably occur in co-simulation of virtual and real-time systems, and which influence system behavior in an unknown and generally detrimental way. This contribution introduces a novel, adaptive method to compensate for constant time-delays in potentially highly nonlinear, spatially distributed mixed real-virtual prototypes, using small feedforward neural networks. Their optimal initialization with respect to defined frequency domain features results from a-priori frequency domain analysis of the entire coupled system, including coupling faults and compensation methods. A linear and a nonlinear example demonstrate the method and emphasize its suitability for nonlinear systems due to online training and adaptation. As the compensation method requires knowledge only of the bandwidths, the proposed method is applicable to distributed mixed real-virtual prototypes in general.

eess.SY↗