A Framework for Graph-Conditioned Hierarchical Shapley Attribution in Patent Valuation

· Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Intellectual Property & Patents · Depth: Expert, quick

Summary

PatentXAI is a novel framework addressing the complex problem of estimating a single patent's economic contribution within products containing thousands of patents. It redefines patent valuation as an explainable AI challenge, employing Shapley values to fairly allocate product profit based on efficiency, symmetry, dummy, and additivity principles. To ensure computational tractability, PatentXAI restricts each patent's coalition to its Markov Blanket within a knowledge graph, leveraging the C-SVE conditional independence theorem. Scaling experiments with n=12 to n=100 patents, using Pareto-distributed coverage graphs, demonstrated a median Markov Blanket size of 32.9 percent of n at n=100, with a 90th-percentile size of 55.2 percent of n, and a runtime of 10 milliseconds per patent. Accuracy metrics showed a 0.088 difference against exact ground truth at n=12 and 0.062 ± 0.003 against a Monte Carlo reference at n=100. The framework also implements a hierarchical profit allocation strategy.

Key takeaway

For intellectual property economists or legal professionals valuing complex patent portfolios, PatentXAI offers a robust, computationally tractable framework. You can apply this method to objectively determine individual patent contributions to product profit, moving beyond traditional estimation challenges. This approach provides a defensible basis for licensing negotiations or litigation, enabling more precise financial assessments of intellectual assets.

Key insights

PatentXAI uses graph-conditioned hierarchical Shapley attribution to fairly value individual patents within large portfolios, treating it as an explainable AI problem.

Principles

Method

PatentXAI computes Shapley values for patent subsets restricted to Markov Blankets in a knowledge graph, then hierarchically allocates profit: exact Shapley for macro-components, centrality-weighted for covering patents.

In practice

Topics

Best for: AI Scientist, Research Scientist, Legal Professional

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.