Extracting entities and relationships from unstructured text is crucial for knowledge graph construction. However, existing works in solving the overlapping relationship and nested entity problems of Chinese text is few. Furthermore, most previous works based on the joint method cannot identify entity types. In terms of these problems, we introduce a joint model of Chinese entity-relation extraction based on a novel pointer cascade tagging strategy (NPCTS). The model used the pointer tagging strategy based on entity type to solve the problem of nested entity and to predict the entity type. It also models relations as functions that map subjects to objects in a sentence, which naturally handles the overlapping problem. Experiments show that the model has outstanding performance on Chinese datasets in two different fields, achieving F1 values of 81.1% and 58.4% respectively. NPCTS can effectively cope with problems of nested entity and overlapping relationship. The entity-relation tuples extracted by this model can be directly imported into graph database, which greatly simplified the process of constructing knowledge graph from unstructured text.
NPCTS模型是针对中文文本的实体关系抽取任务,比较难找到统一对比的数据集与经典的算法。为了进一步验证模型性能,体现本文方法在准确率和稳定性上的优势,将NPCTS模型及其变种与Bert+Bi-LSTM、Bert+Bi-LSTM+CRF、CopyRRL[16]和CalRel[11]四种模型在不同语料库上进行实验对比。Bert+Bi-LSTM为流水线学习的实体关系抽取模型,可以完成命名实体识别任务,并在此基础上判别实体对的关系类型;Bert+Bi-LSTM+CRF(conditional random field)是使用较为普遍的命名实体识别模型,采用CRF分类器识别实体类型。CopyRRL[16]与CasRel[11]模型在英文数据集NYT与WebNLG上都取得了不错的结果,但无法在提取三元组的同时判别头实体与尾实体的类型。本文使用上述模型在中文数据集上复现,实验结果如表2所示,表中“-”代表该模型无法完成该项任务。
QINC J, HOUH Q. Mapping knowledge domain—A new field of information management and knowledge management [J].Journal of Academic Libraries, 2009,27(1):30-37. DOI: 10.3969/j.issn.1002-1027.2009.01.007(Ch ).
[4]
HASEGAWAT, SEKINES, GRISHMANR. Discovering relations among named entities from large corpora [C]//Proceedings of the 42nd Annual Meeting on Association for Computational Linguistics. Morristown: Association for Computational Linguistics, 2004: 415-422. DOI:10.3115/1218955.1219008 .
HUANGC, QIANL H, ZHOUG D, et al. Research on unsupervised Chinese entity relation extraction based on convolution tree kernel [J]. Journal of Chinese Information Processing, 2010, 24(4): 11-17(Ch). DOI:10.3969/j.issn.1003-0077.2010.04.002(Ch ).
[7]
MINTZM, BILLSS, SNOWR, et al. Distant supervision for relation extraction without labeled data [C] //Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP. Morristown:Association for Computational Linguistics,2009:1003-1011. DOI:10.3115/1690219.1690287 .
GANL X, WANC X, LIUD X, et al. Chinese named entity relation extraction based on syntactic and semantic features [J]. Journal of Computer Research and Development, 2016, 53(2): 284-302 (Ch). DOI: 10.7544/issn1000-1239.2016.20150842 .
[10]
ZENGD, LIUK, LAIS, et al. Relation classification via convolutional deep neural network[C] //Proceedings of COLING 2014, the 25th International Conference on Computational Linguistics. Heidelberg:Springer-Verlag,2014:2335-2344.
SUNJ D, GUX S, LIY, et al. Chinese entity relation extraction algorithms based on COAE2016 datasets [J]. Journal of Shandong University (Natural Science), 2017, 52(9): 7-12. DOI:10.6040/j.issn.1671-9352.1.2016.PC7(Ch ).
[13]
SOCHERR, HUVALB, MANNINGC D, et al. Semantic compositionality through recursive matrix-vector spaces [C] //Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning. New York:Association for Computational Linguistics,2012:1201-1211.
[14]
XUY, MOUL L, LIG, et al. Classifying relations via long short term memory networks along shortest dependency paths [C]//Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: Association for Computational Linguistics,2015:1785-1794. DOI:10.18653/v1/d15-1206 .
[15]
WEIZ P, SUJ L, WANGY, et al. A novel cascade binary tagging framework for relational triple extraction[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg: Association for Computational Linguistics, 2020: 1476-1488. DOI:10.18653/v1/2020.acl-main.136 .
[16]
MIWAM, BANSALM. End-to-end relation extraction using LSTMs on sequences and tree structures [C]//Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Stroudsburg: Association for Computational Linguistics, 2016: 1105-1116. DOI:10.18653/v1/p16-1105 .
[17]
ZHENGS C, WANGF, BAOH Y, et al. Joint extraction of entities and relations based on a novel tagging scheme [C]//Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Stroudsburg: Association for Computational Linguistics, 2017: 1227-1236. DOI:10.18653/v1/p17-1113 .
[18]
WEBBERJ. A programmatic introduction to Neo4j [C]// Proceedings of the 3rd Annual Conference on Systems, Programming, and Applications: Software for Humanity. New York: ACM Press, 2012: 217-218. DOI:10.1145/2384716.2384777 .
[19]
DEVLINJ, CHANGM W, LEEK, et al. Bert: Pre-training of Deep Bidirectional Transformers for Language Understanding [DB/OL]. [2021-05-06].DOI: 10.18653/v1/n19-1423 .
[20]
ZENGX R, HES Z, ZENGD J, et al. Learning the extraction order of multiple relational facts in a sentence with reinforcement learning [C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP⁃IJCNLP). Stroudsburg: Association for Computational Linguistics, 2019: 367-377. DOI:10.18653/v1/d19-1035 .