Open-world Semi-supervised Generalized Relation Discovery Aligned in a Real-world Setting

Open-world Relation Extraction (OpenRE) has recently garnered significant attention. However, existing approaches tend to oversimplify the problem by assuming that all instances of unlabeled data belong to novel classes, thereby limiting the practicality of these methods. We argue that the OpenRE setting should be more aligned with the characteristics of real-world data. Specifically, we propose two key improvements: (a) unlabeled data should encompass known and novel classes, including negative instances; and (b) the set of novel classes should represent long-tail relation types. Furthermore, we observe that popular relations can often be implicitly inferred through specific patterns, while long-tail relations tend to be explicitly expressed. Motivated by these insights, we present a method called KNoRD (Known and Novel Relation Discovery), which effectively classifies explicitly and implicitly expressed relations from known and novel classes within unlabeled data. Experimental evaluations on several Open-world RE benchmarks demonstrate that KNoRD consistently outperforms existing methods, achieving significant gains. Prompt model inference Prompt model training • In 1953, five years after the state was established, the JNF was dissolved. JNF [mask] 1953 → dissolved • Overgrowth of escherichia coli causes IBS symptoms. escherichia coli [mask] IBS → causes • The parade took place in Philadelphia, Pennsylvania. Philadelphia [mask] Pennsylvania → located in Clustering + majorityvote bifurcation Constrained, in-sentence predictions (explicit relations) known & novel classes Labeled data Unlabeled data Language Model (DeBERTa) Cross-entropy Loss ? ? Prompt template: sentence + head entity + [mask] + tail entity → predicted relation class Unconstrained, all vocabulary predictions (implicit relations) known classes only Frozen model weights Sample input: Bill graduated with a Master's degree from Auburn University. Bill [mask] Auburn University graduated, degree, Master's attended, represented, member Gold Labels Weak Labels