面向知识图谱的二阶段复杂问句生成框架

张琨; 王元卓; 仇韫琦; 白龙; 江旭晖; 侯坤; 岑建何; 沈华伟; 程学旗

doi:10.12263/DZXB.20240331

您当前的位置：

首页 >

文章列表页 >

面向知识图谱的二阶段复杂问句生成框架

学术论文 | 更新时间：2025-10-16

- 面向知识图谱的二阶段复杂问句生成框架
- A Two-Stage Framework for Complex Question Generation over Knowledge Graph
- 电子学报 2025年53卷第6期页码：2104-2117
- 作者机构：
  
  1.中国科学院计算技术研究所数据智能系统研究中心，北京 100190
  2.中国科学院大学计算机科学与技术学院，北京 101408
  3.中科大数据研究院，河南郑州 450046
  4.中国科学院计算技术研究所网络数据科学与技术重点实验室，北京 100190
  5.北京工商大学计算机与人工智能学院，北京 100048
- 作者简介：
  
  [ "张琨男，1997年生.现为中国科学院计算技术研究所博士研究生.主要研究方向为知识库问答、问句生成. E-mail: zhangkun18z@ict.ac.cn" ]
  [ "王元卓男，1978年生.现为中国科学院计算技术研究所研究员，博士生导师，中科大数据研究院院长.主要研究方向为网络大数据分析、开放知识计算、社交网络演化计算. E-mail: wangyuanzhuo@ict.ac.cn" ]
- 基金信息：
  
  河南省重点研发专项(241111211900);国家自然科学基金(62172393;U1836206;U21B2046);河南省中原人才计划(204200510002)
- DOI：10.12263/DZXB.20240331
  中图分类号： TP391;TP182
- 收稿：2024-04-15，
  
  修回：2024-11-15，
  
  纸质出版：2025-06-25
- 稿件说明：
移动端阅览
张琨, 王元卓, 仇韫琦, 等. 面向知识图谱的二阶段复杂问句生成框架[J]. 电子学报, 2025, 53(06): 2104-2117.

ZHANG Kun, WANG Yuan-zhuo, QIU Yun-qi, et al. A Two-Stage Framework for Complex Question Generation over Knowledge Graph[J]. Acta Electronica Sinica, 2025, 53(06): 2104-2117.
张琨, 王元卓, 仇韫琦, 等. 面向知识图谱的二阶段复杂问句生成框架[J]. 电子学报, 2025, 53(06): 2104-2117. DOI：10.12263/DZXB.20240331

ZHANG Kun, WANG Yuan-zhuo, QIU Yun-qi, et al. A Two-Stage Framework for Complex Question Generation over Knowledge Graph[J]. Acta Electronica Sinica, 2025, 53(06): 2104-2117. DOI：10.12263/DZXB.20240331

摘要

面向知识图谱的问句生成（Question Generation over Knowledge Graph，KGQG）任务是根据知识图谱（Knowledge Graph，KG）子图生成自然语言问句.现有方法通常是直接将实例化的KG子图转换为问句，并且大多采用教师强制（Teacher-Forcing）的训练策略.然而，当前方法仍然面临两个主要挑战：（1）实例化的KG子图缺乏确定性查询意图的整合，导致输入与目标输出之间存在语义歧义现象；（2）采用教师强制训练策略训练的生成模型在推理阶段存在曝光偏差问题.为了缓解语义歧义带来的挑战，本文提出了一个复杂问句生成框架，其包括两个阶段，即事实-查询和查询-问句生成阶段.在第一阶段，本文设计了一个查询图生成器，将KG子图转换为具有不同查询意图的查询图.在第二阶段，本文提出了一个问句生成模型，该模型利用密集连接图卷积网络（Densely Connected Graph Convolutional Network，DCGCN）对查询图进行编码，并利用双向自回归变换器（Bidirectional and Auto-Regressive Transformers，BART）模型进行解码以生成问句.此外，为了减轻曝光偏差问题，本文引入了生成对抗模仿学习对问句生成模型进行训练.其中，所采用的判别器通过模仿标记数据自适应地学习奖励函数，并指导问句生成模型探索潜在问题空间中的高奖励区域.本文在三个广泛使用的数据集上进行了大量实验，结果表明所提出的框架具有显著的有效性.

Abstract

Question generation over knowledge graph (KGQG) aims to generate natural language questions from knowledge graph (KG) facts automatically. Existing methods directly transform an instantiated KG subgraph into a question and usually adopt the teacher-forcing training strategy. However

the current methods still face two major challenges: (1) instantiated KG subgraphs lack the integration of deterministic query intention

resulting in a semantic mismatch between the input and the target output; (2) the teacher-forcing training strategy suffers from exposure bias in the inference stage. To address the challenges posed by semantic ambiguity

this paper proposes a framework for complex question generation consisting of two stages

namely

facts-to-query and query-to-question. In the first stage

this paper designs a query graph generator

which converts KG subgraphs into query graphs with different query intentions. In the second stage

this paper proposes a question generation model

which employs densely connected graph convolutional networks (GCN) to encode the query graphs and utilizes the bidirectional and auto-regressive transformers (BART) model for decoding to generate questions. Moreover

to alleviate exposure bias

we train the question generator with generative adversarial imitation learning. The adopted discriminator learns reward functions self-adaptively through imitating the labeled data and guides the question generator to explore the high-reward area of the potential question space. Extensive experiments conducted on three widely-used datasets demonstrate the significant effectiveness of the proposed framework.

关键词

Keywords

references

BI S , CHENG X Y , LI Y F , et al . Knowledge-enriched, type-constrained and grammar-guided question generation over knowledge bases [EB/OL ] . ( 2020-10-23 )[ 2024-04-25 ] . https://arxiv.org/abs/2010.03157v3 https://arxiv.org/abs/2010.03157v3 .

ELSAHAR H , GRAVIER C , LAFOREST F . Zero-shot question generation from knowledge graphs for unseen predicates and entity types [EB/OL ] . ( 2018-02-19 )[ 2024-04-25 ] . http://arxiv.org/abs/1802.06842 http://arxiv.org/abs/1802.06842 .

INDURTHI S R , RAGHU D , KHAPRA M M , et al . Generating natural language question-answer pairs from a knowledge graph using a RNN based question generation model [C ] // Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1 , Long Papers . Valencia : ACL , 2017 : 376 - 385 .

KUMAR V , HUA Y C , RAMAKRISHNAN G , et al . Difficulty-Controllable Multi-Hop Question Generation from Knowledge Graphs [M ] // The Semantic Web-ISWC 2019 . Cham : Springer International Publishing , 2019 : 382 - 398 .

LIU C , LIU K , HE S Z , et al . Generating questions for knowledge bases via incorporating diversified contexts and answer-aware loss [EB/OL ] . ( 2019-10-29 )[ 2024-04-25 ] . https://arxiv.org/abs/1910.13108v1 https://arxiv.org/abs/1910.13108v1 .

RANZATO M , CHOPRA S , AULI M , et al . Sequence level training with recurrent neural networks [EB/OL ] . ( 2016-05-06 )[ 2024-04-25 ] . https://arxiv.org/abs/1511.06732v7 https://arxiv.org/abs/1511.06732v7 .

GUO Z J , ZHANG Y , TENG Z Y , et al . Densely connected graph convolutional networks for graph-to-sequence learning [J ] . Transactions of the Association for Computational Linguistics , 2019 , 7 : 297 - 312 .

LEWIS M , LIU Y H , GOYAL N , et al . BART: Denoising sequence-to-sequence pre-training for natural language generation, translation, and comprehension [EB/OL ] . ( 2019-10-29 )[ 2024-04-25 ] . https://arxiv.org/abs/1910.13461v1.17 https://arxiv.org/abs/1910.13461v1.17 .

HO J , ERMON S , HO J , et al . Generative adversarial imitation learning [C ] // Proceedings of the 30th International Conference on Neural Information Processing Systems . New York : ACM , 2016 : 4572 - 4580 .

YIH W T , CHANG M W , HE X D , et al . Semantic parsing via staged query graph generation: Question answering with knowledge base [C/OL ] . ( 2015-05-01 )[ 2024-04-05 ] . https://www.microsoft.com/en-us/research/publication/semantic-parsing-via-staged-query-graph-generation-question-answering-with-knowledge-base/ https://www.microsoft.com/en-us/research/publication/semantic-parsing-via-staged-query-graph-generation-question-answering-with-knowledge-base/ .

QIU Y Q , ZHANG K , WANG Y Z , et al . Hierarchical query graph generation for complex question answering over knowledge graph [C ] // Proceedings of the 29th ACM International Conference on Information Knowledge Management . New York : ACM , 2020 : 1285 - 1294 .

HOMMEAUX E P . SPARQL query language for RDF [J/OL ] . ( 2013-03-26 )[ 2024-04-05 ] . http://www.w3.org/TR/rdf-sparql-query/ http://www.w3.org/TR/rdf-sparql-query/ .

仇韫琦 , 王元卓 , 白龙 , 等 . 面向知识库问答的问句语义解析研究综述 [J ] . 电子学报 , 2022 , 50 ( 9 ): 2242 - 2264 .

QIU Y Q , WANG Y Z , BAI L , et al . A survey of question semantic parsing for knowledge base question answering [J ] . Acta Electronica Sinica , 2022 , 50 ( 9 ): 2242 - 2264 . (in Chinese)

张元鸣 , 姬琦 , 徐雪松 , 等 . 基于知识图谱关系路径的多跳智能问答模型研究 [J ] . 电子学报 , 2023 , 51 ( 11 ): 3092 - 3099 .

ZHANG Y M , JI Q , XU X S , et al . Knowledge graph relation path network for multi-hop intelligent question answering [J ] . Acta Electronica Sinica , 2023 , 51 ( 11 ): 3092 - 3099 . (in Chinese)

高留杰 , 赵文 , 张君福 , 等 . G2S: 基于语义块的知识图谱问答语义解析 [J ] . 电子学报 , 2021 , 49 ( 6 ): 1132 - 1141 .

GAO L J , ZHAO W , ZHANG J F , et al . G2S: Semantic segment based semantic parsing for question answering over knowledge graph [J ] . Acta Electronica Sinica , 2021 , 49 ( 6 ): 1132 - 1141 . (in Chinese)

LI Z , WANG Z J , TAN M , et al . DQ-BART: Efficient sequence-to-sequence model via joint distillation and quantization [EB/OL ] . ( 2022-03-21 )[ 2024-04-05 ] . https://arxiv.org/abs/2203.11239v1 https://arxiv.org/abs/2203.11239v1 .

LEVI F W . Ordered groups [C ] // Proceedings of the Joint Conference of the 53rd Annual Meeting of the ACL and the 7th International Joint Conference on Natural Language Processing of the AFNLP . Beijing : Association for Computational Linguistics , 2015 : 15 .

BECK D , HAFFARI G , COHN T . Graph-to-sequence learning using gated graph neural networks [EB/OL ] . ( 2018-06-26 )[ 2024-04-25 ] . https://arxiv.org/abs/1806.09835v1 https://arxiv.org/abs/1806.09835v1 .

BAHDANAU D , CHO K , BENGIO Y . Neural Machine Translation by Jointly Learning to Align and Translate [M/OL ] . ( 2016-05-19 )[ 2024-04-05 ] . http://arxiv.org/abs/1409.0473 http://arxiv.org/abs/1409.0473 .

TU Z P , LU Z D , LIU Y , et al . Modeling coverage for neural machine translation [EB/OL ] . ( 2016-08-06 )[ 2024-04-25 ] . https://arxiv.org/abs/1601.04811v6 https://arxiv.org/abs/1601.04811v6 .

GU J T , LU Z D , LI H , et al . Incorporating copying mechanism in sequence-to-sequence learning [EB/OL ] . ( 2016-06-08 )[ 2024-04-25 ] . https://arxiv.org/abs/1603.06393v3 https://arxiv.org/abs/1603.06393v3 .

WILLIAMS R J . Simple statistical gradient-following algorithms for connectionist reinforcement learning [J ] . Machine Learning , 1992 , 8 ( 3 ): 229 - 256 .

DEVLIN J , CHANG M W , LEE K , et al . BERT: Pre-training of deep bidirectional transformers for language understanding [EB/OL ] . ( 2019-05-24 )[ 2024-04-25 ] . https://arxiv.org/abs/1810.04805v2 https://arxiv.org/abs/1810.04805v2 .

YIH W T , RICHARDSON M , MEEK C , et al . The value of semantic parse labeling for knowledge base question answering [C ] // Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) . Redmond : Microsoft Research , 2016 : 201 - 206 .

TALMOR A , BERANT J . The web as a knowledge-base for answering complex questions [EB/OL ] . ( 2018-03-18 )[ 2024-04-25 ] . https://arxiv.org/abs/1803.06643v1 https://arxiv.org/abs/1803.06643v1 .

RAFFEL C , SHAZEER N , ROBERTS A , et al . Exploring the limits of transfer learning with a unified text-to-text transformer [J ] . Journal of Machine Learning Research , 2020 , 21 ( 1 ): 5485 - 5551 .

KE P , JI H Z , RAN Y , et al . JointGT: Graph-text joint representation learning for text generation from knowledge graphs [EB/OL ] . ( 2021-06-19 )[ 2024-04-05 ] . https://arxiv.org/abs/2106.10502v1 https://arxiv.org/abs/2106.10502v1 .

GUO S S , ZHANG J , WANG Y L , et al . DSM: Question generation over knowledge base via modeling diverse subgraphs with meta-learner [C ] // Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing . Abu Dhabi : Association for Computational Linguistics , 2022 : 4194 - 4207 .

ZHOU M T , HUANG M L , ZHU X Y . An interpretable reasoning network for multi-relation question answering [EB/OL ] . ( 2018-06-01 )[ 2024-04-25 ] . https://arxiv.org/abs/1801.04726v3 https://arxiv.org/abs/1801.04726v3 .

BOLLACKER K , EVANS C , PARITOSH P , et al . Freebase: A collaboratively created graph database for structuring human knowledge [C ] // Proceedings of the 2008 ACM SIGMOD International Conference on Management of Data . New York : ACM , 2008 : 1247 - 1250 .

DU X Y , SHAO J R , CARDIE C . Learning to ask: Neural question generation for reading comprehension [EB/OL ] . ( 2017-04-29 )[ 2024-04-25 ] . https://arxiv.org/abs/1705.00106v1 https://arxiv.org/abs/1705.00106v1 .

HAN X , CAO S L , LV X , et al . OpenKE: An open toolkit for knowledge embedding [C ] // Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing: System Demonstrations . Brussels : Association for Computational Linguistics , 2018 : 139 - 144 .

BORDES A , USUNIER N , GARCIA-DURÁN A , et al . Translating embeddings for modeling multi-relational data [C ] // Proceedings of the 27th International Conference on Neural Information Processing Systems - Volume 2 . New York : ACM , 2013 : 2787 - 2795 .

PENNINGTON J , SOCHER R , MANNING C . Glove: Global vectors for word representation [C ] // Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) . Stanford : Stanford University , 2014 : 1532 - 1543 .

PAPINENI K , ROUKOS S , WARD T , et al . BLEU: A method for automatic evaluation of machine translation [J ] . Proceedings of the Annual Meeting of the Association for Computational Linguistics , 2002 , 2002 : 311 - 318 .

BANERJEE S , LAVIE A . METEOR: An automatic metric for MT evaluation with improved correlation with human judgments [C ] // Proceedings of the ACL Workshop on Intrinsic and Extrinsic Evaluation Measures for Machine Translation and/or Summarization . Pittsburgh : Carnegie Mellon University , 2005 : 65 - 72 .

LIN C Y . Rouge: A package for automatic evaluation of summaries [C ] // Text Summarization Branches Out . Los Angeles : Association for Computational Linguistics , 2004 : 74 - 81 .

BERANT J , LIANG P . Semantic parsing via paraphrasing [C ] // Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) . Stanford : Stanford University , 2014 : 1415 - 1425 .

JIA R , LIANG P . Data recombination for neural semantic parsing [EB/OL ] . ( 2016-06-11 )[ 2024-04-25 ] . https://arxiv.org/abs/1606.03622v1 https://arxiv.org/abs/1606.03622v1 .

KOČISKÝ T , MELIS G , GREFENSTETTE E , et al . Semantic parsing with semi-supervised sequential autoencoders [EB/OL ] . ( 2016-09-29 )[ 2024-04-25 ] . https://arxiv.org/abs/1609.09315v1 https://arxiv.org/abs/1609.09315v1 .

CHEN Y , WU L F , ZAKI M J . Toward subgraph-guided knowledge graph question generation with graph neural networks [J ] . IEEE Transactions on Neural Networks and Learning Systems , 2024 , 35 ( 9 ): 12706 - 12717 .

GOODFELLOW I , POUGET-ABADIE J , MIRZA M , et al . Generative adversarial networks [J ] . Communications of the ACM , 2020 , 63 ( 11 ): 139 - 144 .

LI R P , CHENG X . DIVINE: A generative adversarial imitation learning framework for knowledge graph reasoning [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) . Hong Kong : Association for Computational Linguistics , 2019 : 2642 - 2651 .

ZHANG T T , JI H , SIL A . Joint entity and event extraction with generative adversarial imitation learning [J ] . Data Intelligence , 2019 , 1 ( 2 ): 99 - 120 .

GUO S S , LIAO L Z , ZHANG J , et al . SGSH: Stimulate large language models with skeleton heuristics for knowledge base question generation [EB/OL ] . ( 2024-04-02 )[ 2024-04-05 ] . https://arxiv.org/abs/2404.01923v1 https://arxiv.org/abs/2404.01923v1 .

HU S , ZOU L , ZHU Z X . How question generation can help question answering over knowledge base [M ] // Natural Language Processing and Chinese Computing . Cham : Springer International Publishing , 2019 : 80 - 92 .

CAO R S , ZHU S , LIU C , et al . Semantic parsing with dual learning [EB/OL ] . ( 2019-07-24 )[ 2024-04-25 ] . https://arxiv.org/abs/1907.05343v2 https://arxiv.org/abs/1907.05343v2 .

SERBAN I , GARCÍA-DURÁN A , GÜLÇEHRE Ç , et al . Generating factoid questions with recurrent neural networks: The 30M factoid question-answer corpus [EB/OL ] . ( 2016-05-29 )[ 2024-04-25 ] . https://arxiv.org/abs/1603.06807v2 https://arxiv.org/abs/1603.06807v2 .

RUSSELL S . Learning agents for uncertain environments (extended abstract) [C ] // Proceedings of the Eleventh Annual Conference on Computational Learning Theory . New York : ACM , 1998 : 101 - 103 .

NG A Y , RUSSELL S . Algorithms for inverse reinforcement learning [C ] // Proceedings of the 17th International Conference on Machine Learning . Stanford : Morgan Kaufmann , 2000 : 663 - 670 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于内容语义感知多模态融合的图像增强方法

基于内在质量约束的文本生成和评价综述