Joint entity recognition and relation extraction as a multi-head selection problem

By Giannis Bekoulis, Johannes Deleu, Thomas Demeester, Chris Develder. Expert systems with applications, 2018

405 citations recorded by the source corpus.

Abstract State-of-the-art models for joint entity recognition and relation extraction strongly rely on external natural language processing (NLP) tools such as POS (part-of-speech) taggers and dependency parsers. Thus, the performance of such joint models depends on the quality of the features obtained from these NLP tools. However, these features are not always accurate for various languages and contexts. In this paper, we propose a joint neural model which performs entity recognition and relation extraction simultaneously, without the need of any manually extracted features or the use of any external tool. Specifically, we model the entity recognition task using a CRF (Conditional Random Fields) layer and the relation extraction task as a multi-head selection problem (i.e., potentially identify multiple relations for each entity). We present an extensive experimental setup, to demonstrate the effectiveness of our method using datasets from various contexts (i.e., news, biomedical, real estate) and languages (i.e., English, Dutch). Our model outperforms the previous neural models that use automatically extracted features, while it performs within a reasonable margin of feature-based neural models, or even beats them.

Subfields: Artificial Intelligence, Molecular Biology

View Paper: https://www.semanticscholar.org/paper/6bfe831255162236b7c1f151d7677861190a0472

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Victor Sanh
2. Sebastian Ruder
3. Thomas Wolf
4. Hao Peng
5. Ciyuan Peng
6. Xindong Wu
7. Adrian Ulges
8. Lingfeng Zhong

In Molecular Biology

1. Ciyuan Peng
2. Xindong Wu
3. Adrian Ulges
4. Hao Peng
5. Francesco Osborne
6. Tingwen Liu
7. Markus Eberts
8. Qian Li

Citations in the corpus

• Knowledge Graphs: Opportunities and Challenges (2023, background): “For example, Bekoulis et al. (2018) proposed a joint neural model for cross-lingual (English and Dutch) entity and relation extraction.”
• TPLinker: Single-stage Joint Extraction of Entities and Relations Through Token Pair Linking (2020, methodology): “This model can only copy the last word of an entity; (3) MultiHead (Bekoulis et al., 2018) first identifies all candidate entities, then formulates the task as a multi-head selection problem; (4) GraphRel (Fu et al., 2019) utilizes graph convolutional network to extract overlapping relations by…”
• A Hierarchical Multi-task Approach for Learning Embeddings from Semantic Tasks (2018, methodology): “2017) and other previous work (Katiyar and Cardie 2017; Bekoulis et al. 2018; Augenstein, Ruder, and Søgaard 2018), we do not learn label embeddings, meaning that the (supervised) output/prediction of a layer is not directly fed to the following layer through an embedding learned during training.”
• A Comprehensive Survey on Automatic Knowledge Graph Construction (2023, background): “[136] devise a multi-head selection mechanism to explore all entity/relation combinations.”
• Span-based Joint Entity and Relation Extraction with Transformer Pre-training (2019, methodology): “Bekoulis et al. [3, 2] also employ a bidirectional LSTM to encode each word of the sentence.”
• Structured information extraction from scientific text with large language models (2024, background): “Until recently, there has been relatively little work on relation extraction in materials sciencetext, but there hasbeen much research interestinRE on general-purpose text, especially related to linking people, organizations, locations, and dates 20,21 .”