Probabilistic Circuits for Knowledge Graph Completion with Reduced Rule Sets
Jaikrishna Manojkumar Patil, Nathaniel Lee, Al Mehdi Saadat Chowdhury, YooJung Choi, Paulo Shakarian · 2025
Read Probabilistic Circuits for Knowledge Graph Completion with Reduced Rule Sets by Jaikrishna Manojkumar Patil, Nathaniel Lee, Al Mehdi Saadat Chowdhury, YooJung Choi, Paulo Shakarian (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-06706.txt.
Description
Jaikrishna Patil and co-authors slim the rulebooks of knowledge graph completion without losing their answers. Rule-based systems explain themselves beautifully but need tens of thousands of rules, so nobody can grasp the whole machine. The team instead learns a probability distribution over cooperating rule sets, represented as a probabilistic circuit.
The payoff is drastic: seventy to ninety-six percent fewer rules reach the baseline's peak performance. At equal tiny rule counts their method beats the baseline by up to thirty-one fold, and their minimal sets keep about ninety-one percent of the full system's peak. Across eight benchmarks the small sets also waste less: fewer idle rules, fewer rules per prediction. The framework rests on Nilsson's probabilistic logic and needs no independence assumptions.
Catalog record
| CREATOR | Jaikrishna Manojkumar Patil, Nathaniel Lee, Al Mehdi Saadat Chowdhury, YooJung Choi, Paulo Shakarian |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 43 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-06706