A More Robust Model to Answer Noisy Questions in KBQA
In practical applications, the raw input to a Knowledge Based Question Answering (KBQA) system may vary in forms, expressions, sources, etc. As a result, the actual input to the system may contain various errors caused by various noise in raw data and processes of transmission, transformation, trans...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10058914/ |
| _version_ | 1811155566837366784 |
|---|---|
| author | Ziming Wang Xirong Xu Xinzi Li Haochen Li Li Zhu Xiaopeng Wei |
| author_facet | Ziming Wang Xirong Xu Xinzi Li Haochen Li Li Zhu Xiaopeng Wei |
| author_sort | Ziming Wang |
| collection | DOAJ |
| description | In practical applications, the raw input to a Knowledge Based Question Answering (KBQA) system may vary in forms, expressions, sources, etc. As a result, the actual input to the system may contain various errors caused by various noise in raw data and processes of transmission, transformation, translation, etc. As a result, it is significant to evaluate and enhance the robustness of a KBQA model to various noisy questions. In this paper, we generate 29 datasets of various noisy questions based on the original SimpleQuestions dataset to evaluate and enhance the robustness of a KBQA model, and propose a model which is more robust to various noisy questions. Compared with traditional methods, the main contribution in this paper is that we propose a method of generating datasets of different noisy questions to evaluate the robustness of a KBQA model, and propose a KBQA model which contains incremental learning and Mask Language Model (MLM) in the question answering process, so that our model is less affected by different kinds of noise in questions and achieves higher accuracies on datasets of different noisy questions, which shows its robustness. Experimental results show that our model achieves an average accuracy of 78.1% on these datasets and outperforms the baseline BERT-based model by an average margin of 5.0% with the similar training cost. In addition, further experiments show that our model is compatible with other pre-trained models such as ALBERT and ELECTRA. |
| first_indexed | 2024-04-10T04:36:42Z |
| format | Article |
| id | doaj.art-a63f73bae3b24562bf33feaed1403d9b |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-10T04:36:42Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-a63f73bae3b24562bf33feaed1403d9b2023-03-10T00:00:39ZengIEEEIEEE Access2169-35362023-01-0111227562276610.1109/ACCESS.2023.325260810058914A More Robust Model to Answer Noisy Questions in KBQAZiming Wang0Xirong Xu1https://orcid.org/0000-0002-7558-3031Xinzi Li2Haochen Li3Li Zhu4Xiaopeng Wei5https://orcid.org/0000-0002-8497-611XSchool of Computer Science and Technology, Dalian University of Technology, Dalian, ChinaSchool of Computer Science and Technology, Dalian University of Technology, Dalian, ChinaSchool of Computer Science and Technology, Dalian University of Technology, Dalian, ChinaSchool of Computer Science and Technology, Dalian University of Technology, Dalian, ChinaSchool of Control Science and Engineering, Dalian University of Technology, Dalian, ChinaSchool of Computer Science and Technology, Dalian University of Technology, Dalian, ChinaIn practical applications, the raw input to a Knowledge Based Question Answering (KBQA) system may vary in forms, expressions, sources, etc. As a result, the actual input to the system may contain various errors caused by various noise in raw data and processes of transmission, transformation, translation, etc. As a result, it is significant to evaluate and enhance the robustness of a KBQA model to various noisy questions. In this paper, we generate 29 datasets of various noisy questions based on the original SimpleQuestions dataset to evaluate and enhance the robustness of a KBQA model, and propose a model which is more robust to various noisy questions. Compared with traditional methods, the main contribution in this paper is that we propose a method of generating datasets of different noisy questions to evaluate the robustness of a KBQA model, and propose a KBQA model which contains incremental learning and Mask Language Model (MLM) in the question answering process, so that our model is less affected by different kinds of noise in questions and achieves higher accuracies on datasets of different noisy questions, which shows its robustness. Experimental results show that our model achieves an average accuracy of 78.1% on these datasets and outperforms the baseline BERT-based model by an average margin of 5.0% with the similar training cost. In addition, further experiments show that our model is compatible with other pre-trained models such as ALBERT and ELECTRA.https://ieeexplore.ieee.org/document/10058914/Incremental learningknowledge base question answeringmachine learningnatural language processingrelation predictionrobustness |
| spellingShingle | Ziming Wang Xirong Xu Xinzi Li Haochen Li Li Zhu Xiaopeng Wei A More Robust Model to Answer Noisy Questions in KBQA IEEE Access Incremental learning knowledge base question answering machine learning natural language processing relation prediction robustness |
| title | A More Robust Model to Answer Noisy Questions in KBQA |
| title_full | A More Robust Model to Answer Noisy Questions in KBQA |
| title_fullStr | A More Robust Model to Answer Noisy Questions in KBQA |
| title_full_unstemmed | A More Robust Model to Answer Noisy Questions in KBQA |
| title_short | A More Robust Model to Answer Noisy Questions in KBQA |
| title_sort | more robust model to answer noisy questions in kbqa |
| topic | Incremental learning knowledge base question answering machine learning natural language processing relation prediction robustness |
| url | https://ieeexplore.ieee.org/document/10058914/ |
| work_keys_str_mv | AT zimingwang amorerobustmodeltoanswernoisyquestionsinkbqa AT xirongxu amorerobustmodeltoanswernoisyquestionsinkbqa AT xinzili amorerobustmodeltoanswernoisyquestionsinkbqa AT haochenli amorerobustmodeltoanswernoisyquestionsinkbqa AT lizhu amorerobustmodeltoanswernoisyquestionsinkbqa AT xiaopengwei amorerobustmodeltoanswernoisyquestionsinkbqa AT zimingwang morerobustmodeltoanswernoisyquestionsinkbqa AT xirongxu morerobustmodeltoanswernoisyquestionsinkbqa AT xinzili morerobustmodeltoanswernoisyquestionsinkbqa AT haochenli morerobustmodeltoanswernoisyquestionsinkbqa AT lizhu morerobustmodeltoanswernoisyquestionsinkbqa AT xiaopengwei morerobustmodeltoanswernoisyquestionsinkbqa |