ORACLE: Time-Dependent Recursive Summary Graphs for Foresight on News Data Using LLMs

2025-12-17

AI Summary

This study addresses the decision-making needs of Finnish universities of applied sciences by tackling the challenge of distilling interpretable, domain-adapted policy and trend insights—aggregated at a weekly granularity—from high-volume, daily news streams. Method: We propose the Time-aware Recursive Summarization Graph (TRSG), a novel two-tier architecture integrating hierarchical clustering with large language model (LLM)-driven summarization. TRSG incorporates PESTEL-based multidimensional classification, domain-adaptive filtering, text embedding, and lightweight change detection to enable versioned news crawling, temporal modeling, and thematic evolution tracking. Contribution/Results: TRSG supports automated weekly updates, theme-level differential aggregation, and auditable trend analysis. Deployed in real-world educational intelligence operations, it significantly enhances efficiency in policy assessment and curriculum planning. The system includes a reproducible evaluation framework, demonstrating robustness and practical utility in operational academic settings.

Technology Category

• Natural Language Processing: Summarization
• Application Domains: Humanities & Computational Social Science
• Data Mining & Knowledge Management: Data Visualization & Summarization

Application Category

• Web Mining and Content Analysis: Normalization, clustering, classification, and summarization of Web text
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

ORACLE turns daily news into week-over-week, decision-ready insights for one of the Finnish University of Applied Sciences. The platform crawls and versions news, applies University-specific relevance filtering, embeds content, classifies items into PESTEL dimensions and builds a concise Time-Dependent Recursive Summary Graph (TRSG): two clustering layers summarized by an LLM and recomputed weekly. A lightweight change detector highlights what is new, removed or changed, then groups differences into themes for PESTEL-aware analysis. We detail the pipeline, discuss concrete design choices that make the system stable in production and present a curriculum-intelligence use case with an evaluation plan.

Problem

• Transforms daily news into weekly decision insights
• Applies PESTEL classification and builds summary graphs
• Detects changes and groups differences for analysis

Innovation

• Crawls and versions daily news for weekly insights
• Builds Time-Dependent Recursive Summary Graph using LLM
• Uses lightweight change detector for PESTEL-aware analysis