TRIZBench
From a problem statement to the few contradictions that really matter.
From the thesisTowards an innovative methodology for knowledge fusion from patent datasets: application to inventive design
Thesis of Nathan Witkowicz, conducted with funding from ArcelorMittal
- Problem addressed
- Building a problem graph by hand takes experts days, and it spreads far from the core of the problem.
- Scientific contribution
- Targets the patents and articles around the central problem, builds the graph of its immediate surroundings and extracts a small set of highly relevant contradictions. The user chooses at every step. This graph-building engine, which turns patents and scientific articles into a graph, is meant to run inside Finder.
Related
Theses and projects1
People3
Research themes2
Publications5
- PatenTEB: A Comprehensive Benchmark and Model Family for Patent Text Embedding2025 · arXiv preprint
- DAPFAM: A Domain-Aware Family-level Dataset to benchmark cross domain patent retrieval2025 · arXiv preprint
- SynCRF: Syntax-Based Conditional Random Field for TRIZ Parameter Minings2024 · 16th International Conference on Agents and Artificial Intelligence, ICAART 2024, 24-26 February, 2024
- Leveraging Information Retrieval Pipelines for Inventive Design: Application in Efficient Lattice Structures Manufacturing2024 · 24th IFIP WG 5.4 International TRIZ Future Conference, TFC 2024
- Multi-domain and Heterogeneous Data Driven Innovative Problem Solving: Towards a Unified Representation Framework2023 · 23th International TRIZ Future Conference (TFC)