RESIN: A Dockerized Schema-Guided Cross-document Cross-lingual Cross-media Information Extraction and Event Tracking System

Wen, Haoyang; Lin, Ying; Lai, Tuan Manh; Pan, Xiaoman; Sha, Li; Lin, Xudong; Zhou, Ben; Li, Manling; Wang, Haoyu; Zhang, Hongming; Yu, Xiaodong; Dong, Alexander; Wang, Zhenhailong; Fung, Yi Ren; Mishra, Piyush; Lyu, Qing; Surís, Dídac; Chen, Brian; Brown, Susan Windisch; Palmer, Martha; Callison-Burch, Chris; Vondrick, Carl; Roth, Dan; Chang, Shih‐Fu; Ji, Heng

doi:10.18653/v1/2021.naacl-demos.16

Cited by 36 publications

(38 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An interesting future direction is to study further how to systematically simplify/compress BERT based on the insights obtained using probing experiments to increase efficiency while maintaining effectiveness. We plan to extend our work to other domains as well as other information extraction tasks Wen et al, 2021;Lai et al, 2021a;Li et al, 2020).…”

Section: Discussionmentioning

confidence: 99%

BERT might be Overkill: A Tiny but Effective Biomedical Entity Linker based on Residual Convolutional Neural Networks

Lai¹,

Zhai

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Biomedical entity linking is the task of linking entity mentions in a biomedical document to referent entities in a knowledge base. Recently, many BERT-based models have been introduced for the task. While these models have achieved competitive results on many datasets, they are computationally expensive and contain about 110M parameters. Little is known about the factors contributing to their impressive performance and whether the overparameterization is needed. In this work, we shed some light on the inner working mechanisms of these large BERT-based models. Through a set of probing experiments, we have found that the entity linking performance only changes slightly when the input word order is shuffled or when the attention scope is limited to a fixed window size. From these observations, we propose an efficient convolutional neural network with residual connections for biomedical entity linking. Because of the sparse connectivity and weight sharing properties, our model has a small number of parameters and is highly efficient. On five public datasets, our model achieves comparable or even better linking accuracy than the stateof-the-art BERT-based models while having about 60 times fewer parameters. 1

show abstract

Section: Discussionmentioning

confidence: 99%

BERT might be Overkill: A Tiny but Effective Biomedical Entity Linker based on Residual Convolutional Neural Networks

Lai¹,

Zhai

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…In future work, we are interested in reducing the computational complexity of our baseline model using compression techniques [26,27,28]. We also plan to extend our work to address the task of event coreference resolution [29,30].…”

Section: Discussionmentioning

confidence: 99%

End-to-end Neural Coreference Resolution Revisited: A Simple yet Effective Baseline

Lai¹,

Bui²,

Kim³

2021

Preprint

View full text Add to dashboard Cite

Since the first end-to-end neural coreference resolution model was introduced, many extensions to the model have been proposed, ranging from using higher-order inference to directly optimizing evaluation metrics using reinforcement learning. Despite improving the coreference resolution performance by a large margin, these extensions add a lot of extra complexity to the original model. Motivated by this observation and the recent advances in pre-trained Transformer language models, we propose a simple yet effective baseline for coreference resolution. Our model is a simplified version of the original neural coreference resolution model, however, it achieves impressive performance, outperforming all recent extended works on the public English OntoNotes benchmark. Our work provides evidence for the necessity of carefully justifying the complexity of existing or newly proposed models, as introducing a conceptual or practical simplification to an existing model can still yield competitive results.

show abstract

“…The traditional event extraction task aims to predict the event types by triggers and then extract their arguments according to the pre-defined schema of the event. Similarly, the visual event extraction can be also divided into two subtasks: 1) to predict the visual event types; and 2) to locate and extract objects in source images or videos as visual arguments [21], [67], [92], [93].…”

Section: Visual Event Extractionmentioning

confidence: 99%

“…[67] aligns the situation graph [94] extracted from an image of an event to the abstract meaning representation graph (AMR graph) [98] representing the semantic structure of the caption of this event in terms of the semantic and categories of cross modal arguments. Many constrains on semantic, event type, event argument role and the consistency of visual and text information are also added into joint extraction [21], [67].…”

Section: Challengesmentioning

confidence: 99%

Multi-Modal Knowledge Graph Construction and Application: A Survey

Zhu¹,

Li²,

Wang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Recent years have witnessed the resurgence of knowledge engineering which is featured by the fast growth of knowledge graphs. However, most of existing knowledge graphs are represented with pure symbols, which hurts the machine's capability to understand the real world. The multi-modalization of knowledge graphs is an inevitable key step towards the realization of human-level machine intelligence. The results of this endeavor are Multi-modal Knowledge Graphs (MMKGs). In this survey on MMKGs constructed by texts and images, we first give definitions of MMKGs, followed with the preliminaries on multi-modal tasks and techniques. We then systematically review the challenges, progresses and opportunities on the construction and application of MMKGs respectively, with detailed analyses of the strength and weakness of different solutions. We finalize this survey with open research problems relevant to MMKGs.

show abstract

RESIN: A Dockerized Schema-Guided Cross-document Cross-lingual Cross-media Information Extraction and Event Tracking System

Cited by 36 publications

References 43 publications

BERT might be Overkill: A Tiny but Effective Biomedical Entity Linker based on Residual Convolutional Neural Networks

BERT might be Overkill: A Tiny but Effective Biomedical Entity Linker based on Residual Convolutional Neural Networks

End-to-end Neural Coreference Resolution Revisited: A Simple yet Effective Baseline

Multi-Modal Knowledge Graph Construction and Application: A Survey

Contact Info

Product

Resources

About