Document-level Relation Extraction as Semantic Segmentation

Zhang, Ningyu; Chen, Xiang; Xie, Xin; Tan, Chuanqi; Chen, Mosha; Huang, Fei; Si, Luo; Chen, Huajun

doi:10.24963/ijcai.2021/551

Cited by 89 publications

(65 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CorefBERT: a pre-trained model was proposed by [12] for word embedding. DocuNet-BERT: [31] proposed a U-shaped segmentation module to capture global information among relational triples. GAIN-GloVe/GAIN-BERT: [13] proposed GAIN, which designed mention graph and entity graph to predict target relations, and make use of GloVe or BERT for word embedding, GCN for representation of the graph.…”

Section: Baseline Modelsmentioning

confidence: 99%

CorefDRE: Document-level Relation Extraction with coreference resolution

Xue¹,

Li²,

Dai³

et al. 2022

Preprint

View full text Add to dashboard Cite

Document-level relation extraction is to extract relation facts from a document consisting of multiple sentences, in which pronoun crossed sentences are a ubiquitous phenomenon against a single sentence. However, most of the previous works focus more on mentions coreference resolution except for pronouns, and rarely pay attention to mentionpronoun coreference and capturing the relations. To represent multi-sentence features by pronouns, we imitate the reading process of humans by leveraging coreference information when dynamically constructing a heterogeneous graph to enhance semantic information. Since the pronoun is notoriously ambiguous in the graph, a mention-pronoun coreference resolution is introduced to calculate the affinity between pronouns and corresponding mentions, and the noise suppression mechanism is proposed to reduce the noise caused by pronouns. Experiments on the public dataset, DocRED, Dialo-gRE and MPDD, show that Coref-aware Doc-level Relation Extraction based on Graph Inference Network outperforms the state-of-the-art 1 .

show abstract

Section: Baseline Modelsmentioning

confidence: 99%

CorefDRE: Document-level Relation Extraction with coreference resolution

Xue¹,

Li²,

Dai³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Zhou et al (2021) argue that the Transformer attentions are able to extract useful contextual features across sentences for DocRE, and they adopt an adaptive threshold for each entity pair. Zhang et al (2021) model DocRE as a semantic segmentation task and predict an entity-level relation matrix to capture local and global information.…”

Section: Related Workmentioning

confidence: 99%

“…For DocRED, we consider additional competing methods: Two Phase , which first predicts whether the entity pair has a relation and then predicts the relation type; LSR (Nan et al, 2020), which constructs the graph by inducing a latent document-level graph; Reconstructor (Xu et al, 2021b), which encourages the model to reconstruct a reasoning path during training; DRN (Xu et al, 2021a), which considers different reasoning skills explicitly and uses graph representation and context representation to model the reasoning skills; ATLOP (Zhou et al, 2021), which aggregates contextual information by the Transformer attentions and adopts an adaptive threshold for different entity pairs; and DocuNet (Zhang et al, 2021), which models DocRE as a semantic segmentation task.…”

Section: Setupmentioning

confidence: 99%

“…Test Ign F1 F1 Ign F1 F1 DocRE Systems with GloVe LSR (Nan et al, 2020) 48.82 55.17 52.15 54.18 Reconstructor (Xu et al, 2021b) 54.25 55.70 53.25 55.13 DRN (Xu et al, 2021a 54.61 56.49 54.35 56.33 BiLSTM (Yao et al, 2019) 48 (Nan et al, 2020) 52.43 59.00 56.97 59.05 Reconstructor (Xu et al, 2021b) 58.13 60.18 57.12 59.45 DRN (Xu et al, 2021a) 59.33 61.39 59.15 61.37 ATLOP (Zhou et al, 2021) 59.22 61.09 59.31 61.30 DocuNet (Zhang et al, 2021) 59.86 61.83 59.93 61.86 BERT base (Ye et al, 2020) 54 Implementation Details.…”

Section: Devmentioning

confidence: 99%

See 1 more Smart Citation

Document-Level Relation Extraction with Sentences Importance Estimation and Focusing

Wang¹,

Chen²,

Mou³

et al. 2022

Preprint

View full text Add to dashboard Cite

Document-level relation extraction (DocRE)aims to determine the relation between two entities from a document of multiple sentences. Recent studies typically represent the entire document by sequence-or graphbased models to predict the relations of all entity pairs. However, we find that such a model is not robust and exhibits bizarre behaviors: it predicts correctly when an entire test document is fed as input, but errs when non-evidence sentences are removed. To this end, we propose a Sentence Importance Estimation and Focusing (SIEF) framework for DocRE, where we design a sentence importance score and a sentence focusing loss, encouraging DocRE models to focus on evidence sentences. Experimental results on two domains show that our SIEF not only improves overall performance, but also makes DocRE models more robust. Moreover, SIEF is a general framework, shown to be effective when combined with a variety of base DocRE models. 1

show abstract

“…To capture longterm and multi-level cascading dependencies, deep learning based techniques (e.g., RNNs [10,21,47,66] and CNNs [55,75]) are incorporated into sequential modeling. DNNs are known to have enticing representation capability and have the natural strength to capture comprehensive relations [76] over different entities (e.g., items, users, interactions). Recently, there are works that explore advanced techniques, e.g., memory networks [53], attention mechanisms [56,79], and graph neural networks [9,26,31,36,81] for sequential recommendation [6,23,29,54,61,67,72].…”

Section: Related Work 21 Sequential Recommendationmentioning

confidence: 99%

CauseRec: Counterfactual User Sequence Synthesis for Sequential Recommendation

Zhang

Yao

Zhao

et al. 2021

Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval

View full text Add to dashboard Cite

Learning user representations based on historical behaviors lies at the core of modern recommender systems. Recent advances in sequential recommenders have convincingly demonstrated high capability in extracting effective user representations from the given behavior sequences. Despite significant progress, we argue that solely modeling the observational behaviors sequences may end up with a brittle and unstable system due to the noisy and sparse nature of user interactions logged. In this paper, we propose to learn accurate and robust user representations, which are required to be less sensitive to (attack on) noisy behaviors and trust more on the indispensable ones, by modeling counterfactual data distribution. Specifically, given an observed behavior sequence, the proposed CauseRec framework identifies dispensable and indispensable concepts at both the fine-grained item level and the abstract interest level. CauseRec conditionally samples user concept sequences from the counterfactual data distributions by replacing dispensable and indispensable concepts within the original concept sequence. With user representations obtained from the synthesized user sequences, CauseRec performs contrastive user representation learning by contrasting the counterfactual with the observational. We conduct extensive experiments on real-world public recommendation benchmarks and justify the effectiveness of CauseRec with multiaspects model analysis. The results demonstrate that the proposed CauseRec outperforms state-of-the-art sequential recommenders by learning accurate and robust user representations . CCS CONCEPTS• Information systems → Recommender systems.

show abstract

Document-level Relation Extraction as Semantic Segmentation

Cited by 89 publications

References 5 publications

CorefDRE: Document-level Relation Extraction with coreference resolution

CorefDRE: Document-level Relation Extraction with coreference resolution

Document-Level Relation Extraction with Sentences Importance Estimation and Focusing

CauseRec: Counterfactual User Sequence Synthesis for Sequential Recommendation

Contact Info

Product

Resources

About