High Quality Embeddings for Horn Logic Reasoning
The paper discusses the development of high-quality embeddings for Horn logic reasoning. It introduces various methods to create numeric representations of logical statements that enhance the efficiency of neural networks in ranking choices made by logical reasoners. The authors conduct experiments to evaluate the effectiveness of these embeddings across different knowledge bases.
- ▪Neural networks can be trained to rank choices made by logical reasoners for more efficient searches.
- ▪The paper introduces methods for creating embeddings that improve downstream results.
- ▪Experiments compare different embeddings to identify characteristics suited for specific reasoning tasks.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.20467 (cs) [Submitted on 19 May 2026] Title:High Quality Embeddings for Horn Logic Reasoning Authors:Yifan Zhang, Yasir White, Dean Clark, Joseph Sanchez, Jevon Lipsey, Ashely Hirst, Jeff Heflin View a PDF of the paper titled High Quality Embeddings for Horn Logic Reasoning, by Yifan Zhang and 6 other authors View PDF HTML (experimental) Abstract:Neural networks can be trained to rank the choices made by logical reasoners, resulting in more efficient searches for answers. A key step in this process is creating useful embeddings, i.e., numeric representations of logical statements. This paper introduces and evaluates several approaches to creating embeddings that result in better downstream results.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
