SE-HCL: Schema Enhanced Hybrid Curriculum Learning for Multi-Turn Text-to-SQL
Existing multi-turn Text-to-SQL approaches, mainly use data in a randomized order when training the model, ignoring the rich structural information contained in the dialog and schema. In this paper, we propose to use curriculum learning (CL) to better leverage the curriculum structure of schema, que...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10433573/ |
| _version_ | 1797243279704588288 |
|---|---|
| author | Yiyun Zhang Sheng'an Zhou Gengsheng Huang |
| author_facet | Yiyun Zhang Sheng'an Zhou Gengsheng Huang |
| author_sort | Yiyun Zhang |
| collection | DOAJ |
| description | Existing multi-turn Text-to-SQL approaches, mainly use data in a randomized order when training the model, ignoring the rich structural information contained in the dialog and schema. In this paper, we propose to use curriculum learning (CL) to better leverage the curriculum structure of schema, query, and dialog for multi-turn question-query pairs. We design a model-agnostic framework named Schema Enhanced Hybrid Curriculum Learning (SE-HCL) for multi-turn Text-to-SQL to help the models gain a full contextual semantic understanding. Concretely, We measure the difficulty of the data from both a structural and model perspective. In terms of data structure, we mainly consider the turns of the question and the complexity of the schema and SQL query. Accordingly, we designed a data course module to dynamically adjust the difficulty of the data based on the convergence of the model and the schema enhancement method we designed. In terms of the model, we propose a scoring module that will judge the difficulty of a problem based on whether the model could solve the question effectively. Finally, we will consider both aspects and design a hybrid curriculum to determine the flow of model training. Our experiments show that our proposed method improves SQL-generated performance over previous state-of-the-art models on SparC and CoSQL, especially for hard and long-turn questions. |
| first_indexed | 2024-04-24T18:52:36Z |
| format | Article |
| id | doaj.art-47a6f795963b4862b5300ec1b9f11ca2 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-24T18:52:36Z |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-47a6f795963b4862b5300ec1b9f11ca22024-03-26T17:47:43ZengIEEEIEEE Access2169-35362024-01-0112399023991210.1109/ACCESS.2024.336552210433573SE-HCL: Schema Enhanced Hybrid Curriculum Learning for Multi-Turn Text-to-SQLYiyun Zhang0https://orcid.org/0009-0008-5885-1111Sheng'an Zhou1Gengsheng Huang2Institute of Electronic Information, Guangdong Vocational College, Guangzhou, ChinaInstitute of Electronic Information, Guangdong Vocational College, Guangzhou, ChinaInstitute of Electronic Information, Guangdong Vocational College, Guangzhou, ChinaExisting multi-turn Text-to-SQL approaches, mainly use data in a randomized order when training the model, ignoring the rich structural information contained in the dialog and schema. In this paper, we propose to use curriculum learning (CL) to better leverage the curriculum structure of schema, query, and dialog for multi-turn question-query pairs. We design a model-agnostic framework named Schema Enhanced Hybrid Curriculum Learning (SE-HCL) for multi-turn Text-to-SQL to help the models gain a full contextual semantic understanding. Concretely, We measure the difficulty of the data from both a structural and model perspective. In terms of data structure, we mainly consider the turns of the question and the complexity of the schema and SQL query. Accordingly, we designed a data course module to dynamically adjust the difficulty of the data based on the convergence of the model and the schema enhancement method we designed. In terms of the model, we propose a scoring module that will judge the difficulty of a problem based on whether the model could solve the question effectively. Finally, we will consider both aspects and design a hybrid curriculum to determine the flow of model training. Our experiments show that our proposed method improves SQL-generated performance over previous state-of-the-art models on SparC and CoSQL, especially for hard and long-turn questions.https://ieeexplore.ieee.org/document/10433573/Natural language processingsemantic parsingmulti-turn text-to-SQLcurriculum learning |
| spellingShingle | Yiyun Zhang Sheng'an Zhou Gengsheng Huang SE-HCL: Schema Enhanced Hybrid Curriculum Learning for Multi-Turn Text-to-SQL IEEE Access Natural language processing semantic parsing multi-turn text-to-SQL curriculum learning |
| title | SE-HCL: Schema Enhanced Hybrid Curriculum Learning for Multi-Turn Text-to-SQL |
| title_full | SE-HCL: Schema Enhanced Hybrid Curriculum Learning for Multi-Turn Text-to-SQL |
| title_fullStr | SE-HCL: Schema Enhanced Hybrid Curriculum Learning for Multi-Turn Text-to-SQL |
| title_full_unstemmed | SE-HCL: Schema Enhanced Hybrid Curriculum Learning for Multi-Turn Text-to-SQL |
| title_short | SE-HCL: Schema Enhanced Hybrid Curriculum Learning for Multi-Turn Text-to-SQL |
| title_sort | se hcl schema enhanced hybrid curriculum learning for multi turn text to sql |
| topic | Natural language processing semantic parsing multi-turn text-to-SQL curriculum learning |
| url | https://ieeexplore.ieee.org/document/10433573/ |
| work_keys_str_mv | AT yiyunzhang sehclschemaenhancedhybridcurriculumlearningformultiturntexttosql AT shenganzhou sehclschemaenhancedhybridcurriculumlearningformultiturntexttosql AT gengshenghuang sehclschemaenhancedhybridcurriculumlearningformultiturntexttosql |