SCANNER: Knowledge-Enhanced Approach for Robust Multi-modal Named Entity Recognition of Unseen Entities

TL;DR

SCANNER tackles unseen-entity generalization and noisy annotations in multi-modal NER by pairing a two-stage span-candidate detection process with a knowledge-augmented recognition stage. It constructs prompts that embed candidates and retrieved knowledge from Wikipedia, image captions, object regions, and CLIP-based similarities, enabling accurate recognition and grounding, even for unseen entities. The model is trained with Trust Your Teacher, a self-distillation scheme that adaptively balances ground-truth supervision and teacher predictions via a reliability-aware weighting mechanism. Empirical results show SCANNER achieving competitive NER performance and superior results in MNER and GMNER, driven by effective knowledge integration and the TYT strategy, establishing a strong baseline for knowledge-grounded, multi-modal NER tasks.

Abstract

Recent advances in named entity recognition (NER) have pushed the boundary of the task to incorporate visual signals, leading to many variants, including multi-modal NER (MNER) or grounded MNER (GMNER). A key challenge to these tasks is that the model should be able to generalize to the entities unseen during the training, and should be able to handle the training samples with noisy annotations. To address this obstacle, we propose SCANNER (Span CANdidate detection and recognition for NER), a model capable of effectively handling all three NER variants. SCANNER is a two-stage structure; we extract entity candidates in the first stage and use it as a query to get knowledge, effectively pulling knowledge from various sources. We can boost our performance by utilizing this entity-centric extracted knowledge to address unseen entities. Furthermore, to tackle the challenges arising from noisy annotations in NER datasets, we introduce a novel self-distillation method, enhancing the robustness and accuracy of our model in processing training data with inherent uncertainties. Our approach demonstrates competitive performance on the NER benchmark and surpasses existing methods on both MNER and GMNER benchmarks. Further analysis shows that the proposed distillation and knowledge utilization methods improve the performance of our model on various benchmarks.

Figures (7)

• Figure 1: Illustrations of NER, MNER, and GMNER tasks. The NER task aims to identify named entities from the given text. MNER extends this task to utilize additional image informations. GMNER additionally requires the model to predict entity bounding boxes in the given image, if they are present.
• Figure 2: 'Kroger' is an unseen entity that is hard to recognize as an Organization or Location. By our knowledge base model, it brings to successful prediction.
• Figure 3: The overall architecture of the proposed SCANNER method. The two-stage structure allows for efficient extraction and utilization of knowledge, as knowledge is extracted only for those entity candidates that were filtered through in stage 1.
• Figure 4: An illustration of the entity recognition module (stage 2). Based on the entity candidates (extracted in stage 1), SCANNER utilizes various knowledge sources such as Wikipedia, image captioner, and object knowledge extractor. The knowledge collected from these sources are then processed by RoBERTa to give the final prediction.
• Figure 5: Experiments of text classification task in MNLI datasets. 'matched' is in-domain, and 'mismatched' is out-domain.