M.A. Belyanova1, G.I. Revunkov2, G.I. Afanasyev3, Yu.E. Gapanyuk4

1-4 Bauman Moscow State Technical University (Moscow, Russia)

1 belyanovama@student.bsmtu.ru; 2 revunkov@bmstu.ru; 3 gaipcs@bmstu.ru; 4 gapyu@bmstu.ru

The task of automatically generating questions from the text belongs to the group of text synthesis tasks.

Classification of ways to generate questions consists of two categories – the classification according to the methods of generating questions and classification according to the input data structure. By generation approaches, the methods for generating questions are divided into logical methods (LOGIC), methods based on machine learning (ML), and hybrid methods (HM).According to the structure of the source data, the methods for generating questions are divided into text data (TEXT), graph data (GRAPH), and hybrid data (HD). Since each classification contains three variants, a total of nine combinations of generation methods and input data structures are possible, but to date, not all of these combinations are “settled” by the corresponding methods.

The proposed architecture is based on the Hybrid Intelligent Information Systems (HIIS) approach. The basic components of the HIIS are the subconsciousness of the information system, the consciousness of the information system, and the boundary model of consciousness and subconsciousness.

The system's subconsciousness includes a text extraction module, a knowledge graph extraction module, a concept and relationships extraction module, a text vectorization module, and a knowledge graph vectorization module.

The system's consciousness includes a module for the logical formation of questions, a module for the formation of questions based on machine learning, a module for logical correction of questions, a module for the hybrid formation of questions, and a quality assessment module.

The proposed HIIS-based approach allows experiments with various architectures of an intelligentquestion generationsystem.

Based on the experiments carried out, we can conclude that the use of metagraph representation of knowledge improves question generation quality.

Belyanova M.A., Revunkov G.I., Afanasyev G.I., Gapanyuk Yu.E. Automatic question generation based on texts and knowledge graphs. Dynamics of complex systems. 2020. T. 14. № 4. Р. 55-64. DOI: 10.18127/j19997493-202004-06 (In Russian).

1. Alsubait T., Parsia B., Sattler U. Ontology-based multiple choice question generation. KI-KünstlicheIntelligenz. 2016. V. 30. № 2. P. 183-188.
2. Mitkov R., Ha L.A. Computer-aided generation of multiple-choice tests. Proceedings of the HLT-NAACL 03 workshop on Building educational applications using natural language processing. V. 2. Association for Computational Linguistics. 2003. Р. 17-22.
3. Stasaski K., Hearst M.A. Multiple choice question generation utilizing an ontology. Proceedings of the 12th Workshop on Innovative Use of NLP for Building Educational Applications. 2017. Р. 303-312.
4. Yue Feng, Yang Wang, Hang Li. A Sequence-to-Sequence Approach to Dialogue State Tracking. arXiv preprint arXiv:2011.09553. 2020.
5. Xinya Du, Junru Shao, Claire Cardie. Learning to Ask: Neural Question Generation for Reading Comprehension. arXiv preprint arXiv:1705.00106. 2017.
6. Papineni K., Roukos S., Ward T., Zhu W.J. BLEU: a method for automatic evaluation of machine translation. Proceedings of the 40th annual meeting on association for computational linguistics. – Association for Computational Linguistics. 2002. Р. 311-318.
7. Chen Y., Wu L., Zaki M.J. Reinforcement learning based graph-to-sequence model for natural question generation. arXiv preprint arXiv:1908.04942. 2019.
8. Heilman M. Automatic Factual Question Generation from Text. PhD Thesis. Carnegie Mellon University. 2011.
9. Beljanova M.A., Kaganov Ju.T., Gapanjuk Ju.E. Arhitektura gibridnoj intellektual'noj informacionnoj sistemy generacii voprosov po tekstu. Trudy VIII Mezhdunar. nauch.-praktich. konf. «Nechetkie sistemy, mjagkie vychislenija i intellektual'nye tehnologii» (NSMVIT-2020). T. 2. S. 35-43 (In Russian).
10. Chernenkiy V.M., Gapanyuk Yu.E., Nardid A.N., Gushcha A.V., Fedorenko Yu.S. The Hybrid Multidimensional-Ontological Data Model Based on Metagraph Approach. In: Petrenko A., Voronkov A. (eds.) Perspectives of System Informatics. PSI 2017. LNCS. Springer, Cham. 2018. V. 10742. Р. 72-87. DOI: https://doi.org/10.1007/978-3-319-74313-4\_6
11. Chernenkiy V., Gapanyuk Yu., Terekhov V., Revunkov G., Kaganov Y. The hybrid intelligent information system approach as the basis for cognitive architecture. Procedia Computer Science. 2018. V. 145. P. 143-152.
12. Tianqiao Liu, Qian Fang, Wenbiao Ding, Zhongqin Wu, Zitao Liu. Mathematical Word Problem Generation from Commonsense Knowledge Graph and Equations. arXiv preprint arXiv:2010.06196. 2020.