Causal knowledge graph construction for enterprise innovation events in the digital economy and its application to strategic decision-making
Abstract The rapid development of the digital economy has created vast streams of textual data on enterprise activities and innovations. Unlocking causal insights from these data is critical for companies seeking strategic advantages. In this paper, we propose a novel framework for constructing a ca...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-06-01
|
| Series: | Journal of King Saud University: Computer and Information Sciences |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44443-025-00086-3 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract The rapid development of the digital economy has created vast streams of textual data on enterprise activities and innovations. Unlocking causal insights from these data is critical for companies seeking strategic advantages. In this paper, we propose a novel framework for constructing a causal knowledge graph of enterprise innovation events mined from digital-economy news and reports. The goal is to represent events (e.g. product launches, strategic investments) as nodes and their causal relationships as directed edges, enabling advanced reasoning about how one innovation leads to another. Our method first extracts enterprise-related events using neural language models, then identifies causal links between events with a specialized classifier, and finally integrates these into a knowledge graph that can support strategic decision-making. We introduce a new architecture that integrates pre-trained language encoders and graph neural networks to jointly model event contexts and global causal structure. The proposed approach is evaluated on three datasets: a DEMAND competition dataset of digital economy innovation events, the FinCausal financial causality dataset, and the Causal News Corpus of news events. Experimental results show that our model outperforms ten state-of-the-art baselines in causal event detection (with F1 improvements up to 3-5% absolute) and produces knowledge graphs that accurately capture event influence patterns. We also conduct ablation studies confirming the contribution of each module. Our findings demonstrate that constructing causal knowledge graphs of innovation events can significantly enhance strategic decision-making, allowing enterprises to predict potential outcomes of innovation strategies and better navigate the digital economy. |
|---|---|
| ISSN: | 1319-1578 2213-1248 |