Barack's Wife Hillary: Using Knowledge-Graphs for Fact-Aware Language Modeling
This addresses the issue of factual accuracy in language models for applications like text generation and question answering, though it is incremental as it builds on existing knowledge graph integration methods.
The paper tackles the problem of traditional language models' inability to recall unseen facts by introducing a knowledge graph language model (KGLM) that selects and copies relevant facts from a knowledge graph, enabling it to generate out-of-vocabulary tokens and achieve significantly better performance than a strong baseline, outperforming even large models in factual sentence completion.
Modeling human language requires the ability to not only generate fluent text but also encode factual knowledge. However, traditional language models are only capable of remembering facts seen at training time, and often have difficulty recalling them. To address this, we introduce the knowledge graph language model (KGLM), a neural language model with mechanisms for selecting and copying facts from a knowledge graph that are relevant to the context. These mechanisms enable the model to render information it has never seen before, as well as generate out-of-vocabulary tokens. We also introduce the Linked WikiText-2 dataset, a corpus of annotated text aligned to the Wikidata knowledge graph whose contents (roughly) match the popular WikiText-2 benchmark. In experiments, we demonstrate that the KGLM achieves significantly better performance than a strong baseline language model. We additionally compare different language model's ability to complete sentences requiring factual knowledge, showing that the KGLM outperforms even very large language models in generating facts.