Inducing Interpretability in Knowledge Graph Embeddings
This addresses the issue of interpretability in knowledge graph embeddings for researchers and practitioners, but it is incremental as it builds on existing Universal Schema methods.
The paper tackles the problem of making knowledge graph embeddings interpretable by proposing a method using entity co-occurrence statistics, which significantly improves interpretability while maintaining comparable performance in other tasks.
We study the problem of inducing interpretability in KG embeddings. Specifically, we explore the Universal Schema (Riedel et al., 2013) and propose a method to induce interpretability. There have been many vector space models proposed for the problem, however, most of these methods don't address the interpretability (semantics) of individual dimensions. In this work, we study this problem and propose a method for inducing interpretability in KG embeddings using entity co-occurrence statistics. The proposed method significantly improves the interpretability, while maintaining comparable performance in other KG tasks.