ORSO QGen: Odds-Ratio Steerable Optimization for Controlling Question Generation
Saved in:
| Title: | ORSO QGen: Odds-Ratio Steerable Optimization for Controlling Question Generation |
|---|---|
| Language: | English |
| Authors: | Andreea Dutulescu, Stefan Ruseti (ORCID |
| Source: | Grantee Submission. 2026. |
| Peer Reviewed: | Y |
| Page Count: | 13 |
| Publication Date: | 2026 |
| Sponsoring Agency: | National Center for Education Research (NCER) (ED/IES) |
| Contract Number: | R305T240035 |
| Document Type: | Speeches/Meeting Papers Reports - Research |
| Descriptors: | Automation, Questioning Techniques, Artificial Intelligence, Technology Uses in Education, Models |
| Abstract: | Question generation plays an important role in educational applications, enabling automated assessment and reading comprehension support. Attribute-controlled question generation aims to produce questions that fit predefined characteristics such as difficulty, focus, or coverage. Existing methods predominantly rely on supervised fine-tuning, which often fails to impose a strong adherence to attribute values, resulting in weak coupling between prompt specifications and model outputs. We introduce Odds-Ratio Steerable Optimization (ORSO), a framework designed to enhance attribute sensitivity in question generation models. Building upon preference-based learning techniques without requiring human-curated preference sets, ORSO uses input-level perturbations to create contrastive training signals. Empirical evaluations on both exhaustive and expert-validated attribute configurations indicate that ORSO performs better than SteerLM and ORPO methods in enforcing attribute conformity while maintaining output quality. These results argue for the benefits of explicit attribute-aware optimization in controllable question generation tasks. [This paper was published in: "Findings of the Association for Computational Linguistics: EACL 2026," Association for Computational Linguistics, 2026, pp. 5248-5259.] |
| Abstractor: | As Provided |
| IES Funded: | Yes |
| Entry Date: | 2026 |
| Access URL: | https://aclanthology.org/2026.findings-eacl.277/ |
| Accession Number: | ED679868 |
| Database: | ERIC |
| Abstract: | Question generation plays an important role in educational applications, enabling automated assessment and reading comprehension support. Attribute-controlled question generation aims to produce questions that fit predefined characteristics such as difficulty, focus, or coverage. Existing methods predominantly rely on supervised fine-tuning, which often fails to impose a strong adherence to attribute values, resulting in weak coupling between prompt specifications and model outputs. We introduce Odds-Ratio Steerable Optimization (ORSO), a framework designed to enhance attribute sensitivity in question generation models. Building upon preference-based learning techniques without requiring human-curated preference sets, ORSO uses input-level perturbations to create contrastive training signals. Empirical evaluations on both exhaustive and expert-validated attribute configurations indicate that ORSO performs better than SteerLM and ORPO methods in enforcing attribute conformity while maintaining output quality. These results argue for the benefits of explicit attribute-aware optimization in controllable question generation tasks. [This paper was published in: "Findings of the Association for Computational Linguistics: EACL 2026," Association for Computational Linguistics, 2026, pp. 5248-5259.] |
|---|
