计算机科学 ›› 2024, Vol. 51 ›› Issue (5): 208-215.doi: 10.11896/jsjkx.230200131

• 人工智能 • 上一篇    下一篇

基于跨层级多视角特征的多语言事件探测

张志远, 张维彦, 宋雨秋, 阮彤   

  1. 华东理工大学信息工程与科学学院 上海 200237
  • 收稿日期:2023-02-19 修回日期:2023-06-21 出版日期:2024-05-15 发布日期:2024-05-08
  • 通讯作者: 阮彤(ruantong@ecust.edu.cn)
  • 作者简介:(1127657064@qq.com)

Multilingual Event Detection Based on Cross-level and Multi-view Features Fusion

ZHANG Zhiyuan, ZHANG Weiyan, SONG Yuqiu, RUAN Tong   

  1. School of Information Science and Engineering,East China University of Science and Technology,Shanghai 200237,China
  • Received:2023-02-19 Revised:2023-06-21 Online:2024-05-15 Published:2024-05-08
  • About author:ZHANG Zhiyuan,born in 1998,postgraduate.His main research interests include natural language processing and multilingual pretraining model.
    RUAN Tong,born in 1973,professor,Ph.D supervisor.Her main research interests include text extraction know-ledge graph and data quality assessment.

摘要: 多语言事件探测任务的目标是将多种语言的新闻文档集合组织成不同的关键事件,其中每个事件可以包含不同语言的新闻文档。该任务有助于各种下游任务应用,如多语言知识图谱构建、事件推理、信息检索等。目前,多语言事件探测主要分为先翻译再事件探测与先单语言检测再跨多种语言对齐两种方法,前者依赖翻译的效果,后者需要为每种语言单独训练模型。为此,提出了一种名为基于跨层级多视角特征融合的多语言事件探测方法,端到端地进行多语言事件探测任务。该方法从不同层级利用文档的多视角特征,获得了高可靠性的多语言事件探测结果并提升了低资源语言事件探测的泛化性能。在9种语言混合的新闻数据集上进行的实验表明,所提方法的BCubed F1值提升了4.63%。

关键词: 多语言预训练模型, 多语言事件探测, 新闻文档聚类, 加权相似度, 增量聚类

Abstract: The goal of the multilingual event detection task is to organize a collection of news documents in multiple languages into different key events,where each event can include news documents in different languages.This task facilitates various downstream task applications,such as multilingual knowledge graph construction,event reasoning,information retrieval,etc.At pre-sent,multilingual event detection is mainly divided into two methods:translation first and then event detection,and single language detection first and then alignment across multiple languages.The former relies on the effect of translation while the latter requires a separate training model for each language.To this end,this paper proposes a multilingual event detection method based on cross-level multi-view feature fusion,which performs end-to-end multilingual event detection tasks.This method uses the multi-view features of documents from different levels to obtain high reliability.It improves the generalization performance of low-resource language event detection.Experiments on a news dataset with a mixture of nine languages show that the proposed method improves the BCubed F1 value by 4.63%.

Key words: Multilingual pre-training model, Multilingual event detection, News documents clustering, Weighted similarity, Incremental clustering

中图分类号: 

  • TP391
[1]HUANG Z,LI Z,JIANG H,et al.Multilingual KnowledgeGraph Completion with Self-Supervised Adaptive Graph Alignment[C]//Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics.2022:474-485.
[2]AHUJA K,KUMAR S,DANDAPAT S,et al.Multi TaskLearning For Zero Shot Performance Prediction of Multilingual Models[C]//Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics.2022:5454-5467.
[3]FUJINUMA Y,JORDAN L,KANN K,et al.Match the Script,Adapt if Multilingual:Analyzing the Effect of Multilingual Pretraining on Cross-lingual Transferability[C]//Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics.2022:1500-1512.
[4]GUZMAN L,LAI V,POURAN A,et al.Event Detection for Suicide Understanding[C]//Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics.2022:1952-1961.
[5]YANG W,BOYD-GRABER J,PHILIP R.A Multilingual Topic Model for Learning Weighted Topic Links Across Corpora with Low Comparability[C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing.2019:1243-1248.
[6]LIU J,CHEN Y,LIU K,et al.Neural Cross-Lingual Event Detection with Minimal Parallel Resources[C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing.2019:738-748.
[7]MIRANDA S,ZNOTINS A,COHEN S,et al.Multilingualclustering of streaming news[C]//Proceedings of the 2018 Conference onEmpirical Methods in Natural Language Processing.2018:4535-4544.
[8]LINGER M,HAJAIEJ M.Batch clustering for multilingualnews streaming[C]//Proceedings of the Text2Story'20 Workshop.2020:55-61.
[9]LABAN P,HEARST M.newsLens:building and visualizinglong-ranging news stories[C]//Proceedings of the Events and Stories in the News Workshop.2017:1-9.
[10]STAYKOVSKI T,BARRON-CEDENO A,MARTINO G,et al.Dense vs.Sparse Representations for News Stream Clustering[C]//Proceedings of the Text2Story'19 Workshop.2019:47-52.
[11]LI Y,YU Z,GAO S,et al.Case-related News Detection Based on Case Element and Deep Clustering Method[J].Journal of Chinese Information Processing,2021,35(11):60-69.
[12]GUO H,WANG Z,ZHU Q,et al.Event Clustering Method for Chinese Social Text Based on Semi-supervised Learning [J].Journal of Chinese Information Processing,2022,36(2):152-159.
[13]SARAVANAKUMAR K,BALLESTEROS M,CHANDRAS-EKARAN M,et al.Event-Driven News Stream Clustering using Entity-Aware Contextual Embeddings[C]//Proceedings of The 16th Conference of the European Chapter of the Association for Computational Linguistics.2021:2330-2340.
[14]DEVLIN J,CHANG M,LEE K,et al.BERT:Pre-training ofDeep Bidirectional Transformers for Language Understanding[C]//Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics.2019:4171-4186.
[15]ZHANG Y,GUO F,SHEN J,et al.Unsupervised Key Event Detection from Massive Text Corpora[C]//Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining.2022:2535-2544.
[16]REIMERS N,GUREVYCH I.Sentence-BERT:Sentence Em-beddings using Siamese BERT-Networks[C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing.2019:3980-3990.
[17]YANG W,YU Z,GAO S,et al.Chinese-Vietnamese news topic discovery methodbased on cross-language neural topic model[J].Journal of Computer Applications,2021,41(10):2879-2884.
[18]CONNEAU A,LAMPLE G,DENOYER L,et al.Word Translation Without Parallel Data[C]//Proceedings of the 6th International Conference on Learning Representations.2018:1-14.
[19]FENG F,YANG Y,CER D,et al.Language-agnostic BERTSentence Embedding[C]//Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics.2022:878-891.
[20]XUE L,CONSTANT N,ROBERTS A,et al.mT5:A Massively Multilingual Pre-trained Text-to-Text Transformer[C]//Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics.2021:483-498.
[21]YANG Z,DAI Z,YANG Y,et al.XLNet:Generalized Auto-regressive Pretraining for Language Understanding[C]//Advances in Neural Information Processing Systems 32.2019:5754-5764.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!