Research Datasets, Models & Financial AI Resources

Over the past 15 years, I have developed a portfolio of datasets, models, and research tools for Natural Language Processing, multilingual AI, and Financial AI. These resources support work on trustworthy AI, financial narrative analysis, Arabic financial NLP, information extraction, document structure understanding, summarisation, and large-scale analysis of corporate reports. Many of these resources have been used in projects involving academic, regulatory, and industry partners, and are made available through GitHub, HuggingFace, and institutional research data repositories to support reproducible and responsible AI research.

FinAraT5

Arabic Financial NLP Generative AI Summarisation

A text-to-text model for Arabic financial language understanding and generation. FinAraT5 supports research on Arabic financial summarisation and domain-specific generative AI for under-resourced financial contexts.

HuggingFace Paper

AraFinNews

Arabic Financial News 212k Articles Financial AI

A large-scale Arabic financial news dataset designed to support research in financial text analytics, domain adaptation, sentiment analysis, topic modelling, summarisation, and Arabic financial language modelling.

GitHub HuggingFace Paper

FinXtract

Financial Reports Document AI Structure Extraction

FinXtract is an LLM-powered tool that transforms annual reports into structured, searchable, and insightful data — with no manual rule-writing.

GitHub Paper

CFIE-FRSE

Financial Reports Document AI Structure Extraction

The Corporate Financial Information Environment, Final Report Structure Extractor, is a desktop application for detecting the structure of UK annual reports and extracting report content at section level. It supports scalable analysis of unstructured financial documents.

GitHub Paper Paper

AraFinNLP 2024

Arabic Financial NLP Shared Task Benchmark Dataset

The first Arabic Financial NLP shared task, introducing benchmark datasets and evaluation tasks for Arabic financial language understanding. The initiative supports research on financial AI, Arabic LLM evaluation, domain adaptation, information extraction, and trustworthy multilingual financial NLP systems.

Project Website Paper

Financial Narrative Summarisation (FNS) Shared Task

Financial Summarisation LLM Evaluation Financial AI Benchmark

A benchmark shared task and evaluation framework for automatic summarisation of financial narratives and long-form corporate disclosures. The dataset and task infrastructure support research on generative AI, transformer-based summarisation, explainability, and financial reporting analysis.

Project Website Paper

FinCausal 2023

Causality Detection Financial Reasoning Explainable AI

A financial document causality detection shared task designed to support research on explainable financial AI and reasoning over corporate disclosures. The benchmark focuses on identifying causal relationships within complex financial narratives and regulatory reporting.

Project Website

Long-Document Financial Narrative Summarisation Evaluation Dataset

Financial NLP Evaluation Metrics Transformer Models

A research benchmark investigating evaluation metrics for long-document financial narrative summarisation using transformer architectures. The work supports robust evaluation practices for trustworthy generative AI and financial summarisation systems.

Paper

CoFiF Plus

French Financial NLP Narrative Summarisation Multilingual Financial AI

A French financial narrative summarisation corpus developed to support multilingual Financial AI research, transformer-based summarisation, and cross-lingual analysis of corporate reporting and financial disclosures.

GitHub Paper

Annual Reports Key Sections Corpora, 2003–2017

Corporate Reporting Narrative Disclosure Financial NLP

Plain text content extracted from 31,464 annual reports published by firms listed on the London Stock Exchange. The corpus supports large-scale research on financial narratives, disclosure, readability, and corporate communication.

DOI: 10.17635/lancaster/researchdata/271

Dataset Paper1 Paper2

UK Annual Report Narratives Dataset

Financial Narratives Textual Features AI for Accounting

A dataset of summary textual features extracted from UK annual report narratives published between 2002 and 2017. It enables research on financial communication, narrative reporting, AI-assisted accounting analysis, and disclosure quality.

DOI: 10.17635/lancaster/researchdata/296

Dataset Paper1 Paper2

Strategic Commentary Dataset

Strategic Reports Corporate Discourse Financial AI

A corpus of strategic commentary extracted from UK annual reports for the period 2003–2018. It supports AI-driven analysis of corporate strategy, business risk, forward-looking statements, and narrative disclosure.

DOI: 10.17635/lancaster/researchdata/261

Dataset
Paper1 Paper2

UK Annual Reports Key Sections

Annual Reports Section-Level Extraction Financial Document AI

A structured dataset of key sections extracted from UK annual reports. It supports research into automatic document segmentation, financial information retrieval, narrative analysis, and AI-based report understanding.

DOI: 10.17635/lancaster/researchdata/262

Dataset Paper1 Paper2

N-gram List for the StratScore Metric

Strategic Disclosure Explainable Metrics Financial Text Analytics

A curated mapping of 709 n-grams used in the StratScore measure to 3,584 n-grams from the existing literature. This resource supports transparent and explainable analysis of strategic language in corporate reporting.

DOI: 10.17635/lancaster/researchdata/232

Paper1 Paper2