Rematch: Robust and Efficient Matching of Local Knowledge Graphs to Improve Structural and Semantic Similarity

Knowledge graphs play a pivotal role in various applications, such as question-answering and fact-checking. Abstract Meaning Representation (AMR) represents text as knowledge graphs. Evaluating the quality of these graphs Figure 1: AMR for the sentence: "He did not cut the involves matching them structurally to each apple with a knife." Colors indicate AMR components: other and semantically to the source text. Existing instances (blue), relations (red), constants (teal), and attributes AMR metrics are inefficient and struggle (orange). The instance cut-01 is a verb frame to capture semantic similarity. We also lack that uses ARG0, ARG1 and inst to express the verb's a systematic evaluation benchmark for assessing agent (he), patient (apple), and instrument (knife), structural similarity between AMR graphs.