XORIJIY YOZMA NUTQDA GRAMMATIK XATOLARNI KLASSIFIKATSIYALASH ALGORITMINI ISHLAB CHIQISH
Article Sidebar
Main Article Content
Abstract
Maqolada xorijiy tildagi yozma matnlarda aniqlangan grammatik xatolarni yirik va mayda lingvistik sinflarga avtomatik ajratish uchun ierarxik va noaniqlikni nazorat qiluvchi GXKA-HC algoritmi taklif etilgan. Algoritm xato segmenti va uning ikki tomonlama kontekstidan Transformer tasvirini, morfologik teglar, sintaktik bog‘lanishlar, tahrir operatsiyasi va topshiriq metama’lumotlarini yagona xususiyatlar vektoriga birlashtiradi. Yirik sinf ehtimoli va shu sinf ichidagi mayda tur ehtimoli shartli ko‘paytma orqali bog‘lanadi; sinflar nomutanosibligi effective-number vazni va focal loss bilan, ierarxik nomuvofiqlik esa KLkonsistensiya hadi bilan cheklanadi. Qaror ishonchliligi entropiya va maksimal posterior marja orqali hisoblanadi; kalibrlash to‘plamida conformal prediction yordamida xato turi bitta yorliq yoki kichik yorliqlar to‘plami ko‘rinishida qaytariladi. Reproduktiv sintetik verifikatsiya 144 virtual profil, 1152 yozma sessiya va 9216 grammatik xato kandidati bilan, profil bo‘yicha ajratilgan olti blokli krossvalidatsiyada bajarildi. To‘liq qamrovda GXKA-HC Macro-F1=0,894±0,006 va MCC=0,889±0,006 ko‘rsatdi; ishonchlilik τ=0,75 bo‘lganda 83,2% avtomatik qamrovda Macro-F1=0,948 ga yetdi. α=0,10 conformal rejimida empirik qamrov 0,901, o‘rtacha yorliqlar to‘plami hajmi 1,032 bo‘ldi. Ushbu qiymatlar real respondentlarning empirik ma’lumoti emas, algoritm va validatsiya protokolining sintetik tekshiruvidir.
Downloads
Article Details
Section
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
How to Cite
References
Bryant C., Yuan Z., Qorib M.R., Cao H., Ng H.T., Briscoe T. Grammatical Error Correction: A Survey of the State of the Art. Computational Linguistics. 2023;49(3):643–701. DOI: 10.1162/coli_a_00478.
Omelianchuk K., Atrasevych V., Chernodub A., Skurzhanskyi O. GECToR – Grammatical Error Correction: Tag, Not Rewrite. BEA 2020. P. 163–170. DOI: 10.18653/v1/2020.bea-1.16.
Bryant C., Felice M., Briscoe T. Automatic Annotation and Evaluation of Error Types for Grammatical Error Correction. ACL 2017. P. 793–805. DOI: 10.18653/v1/P17-1074.
Yang L., Li H., Li L., Xu C., Xia S., Yuan C. LET: Leveraging Error Type Information for Grammatical Error Correction. Findings of ACL 2023. P. 5986–5998. DOI: 10.18653/v1/2023.findings-acl.371.
Fei Y., Cui L., Yang S., Lam W., Lan Z., Shi S. Enhancing Grammatical Error Correction Systems with Explanations. ACL 2023. P. 7489–7501. DOI: 10.18653/v1/2023.acl-long.413.
Uz H., Eryiğit G. Towards Automatic Grammatical Error Type Classification for Turkish. EACL Student Research Workshop 2023. P. 134–142. DOI: 10.18653/v1/2023.eacl-srw.14.
Katinskaia A., Yangarber R. Grammatical Error Correction for Sentencelevel Assessment in Language Learning. BEA 2023. P. 488–502. DOI: 10.18653/v1/2023.bea-1.41.
Omelianchuk K., Liubonko A., Skurzhanskyi O., Chernodub A., Korniienko O., Samokhin I. Pillars of Grammatical Error Correction: Comprehensive Inspection of Contemporary Approaches in the Era of Large Language Models. BEA 2024. DOI: 10.18653/v1/2024.bea-1.3.
Davis C., Caines A., Andersen Ø.E., Taslimipoor S., Yannakoudakis H., Yuan Z., Bryant C., Rei M., Buttery P. Prompting Open-source and Commercial Language Models for Grammatical Error Correction of English Learner Text. Findings of ACL 2024. P. 11952–11967. DOI: 10.18653/v1/2024.findings-acl.711.
Kobayashi M., Mita M., Komachi M. Revisiting Meta-evaluation for Grammatical Error Correction. Transactions of the ACL. 2024;12:837–855. DOI: 10.1162/tacl_a_00676.
Katinskaia A., Yangarber R. GPT-3.5 for Grammatical Error Correction. LREC-COLING 2024. P. 7831–7843.
Chan K.W.H., Bryant C., Nguyen L., Caines A., Yuan Z. Grammatical Error Correction for Code-Switched Sentences by Learners of English. LRECCOLING 2024. P. 7926–7938.
Ye J., Xu Z., Li Y., et al. CLEME2.0: Towards Interpretable Evaluation by Disentangling Edits for Grammatical Error Correction. ACL 2025. P. 204–222. DOI: 10.18653/v1/2025.acl-long.10.
Koyama A., Mita M., Yoon S.-Y., Takama Y., Komachi M. Targeted Syntactic Evaluation for Grammatical Error Correction. ACL 2025. P. 21108–21125. DOI: 10.18653/v1/2025.acl-long.1026.
Cao Y., Wang T., Xu L., Wang Z., Cai M. CxGGEC: Construction-Guided Grammatical Error Correction. ACL 2025. P. 6143–6156. DOI: 10.18653/v1/2025.acl-long.307.
Goto T., Vasselli J., Watanabe T. Improving Explainability of Sentencelevel Metrics via Edit-level Attribution for Grammatical Error Correction. ACL SRW 2025. P. 1004–1015. DOI: 10.18653/v1/2025.acl-srw.77.
Li W., Luo W., Peng G., Wang H. Explanation based In-Context Demonstrations Retrieval for Multilingual Grammatical Error Correction. NAACL 2025. P. 4881–4897. DOI: 10.18653/v1/2025.naacl-long.251.
Palma Gomez F., Rozovskaya A. Low-Resource Grammatical Error Correction: Selective Data Augmentation with Round-Trip Machine Translation. Findings of ACL 2025. P. 25749–25770. DOI: 10.18653/v1/2025.findings-acl.1322.
Staruch R., Graliński F., Dzienisiewicz D. Adapting LLMs for Minimaledit Grammatical Error Correction. BEA 2025. P. 118–128. DOI: 10.18653/v1/2025.bea-1.9.
Gebeşçe A., Şahin G.G. GECTurk WEB: An Explainable Online Platform for Turkish Grammatical Error Detection and Correction. COLING System Demonstrations 2025. P. 163–173.
Akef S., Mendes A., Rebuschat P., Meurers D. LLM-Powered but RuleGrounded: Pedagogically Relevant Grammatical Error Characterization for Learner Model Construction. BEA 2026. P. 851–862. DOI: 10.18653/v1/2026.bea-1.58.
Smirnova A., Kopan A., Makeev V., Chernishev G. What Aggregate Scores Hide: Per-Rule Evaluation of Russian Grammatical Error Correction. BEA 2026. P. 463–478. DOI: 10.18653/v1/2026.bea-1.32.
Goto T., Sakai Y., Watanabe T. Edit-level Majority Voting Mitigates OverCorrection in LLM-based Grammatical Error Correction. BEA 2026. P. 899–913. DOI: 10.18653/v1/2026.bea-1.60.
Östling R., Kurfali M., Caines A. LLM-based Post-editing as Referencefree GEC Evaluation. BEA 2025. P. 213–224. DOI: 10.18653/v1/2025.bea-1.16.
Li J., Wang Q., Zhu C., Mao Z., Zhang Y. Grammatical Error Correction via Mixed-Grained Weighted Training. Findings of EMNLP 2023. P. 6027–6037. DOI: 10.18653/v1/2023.findings-emnlp.400.
Bryant C., Felice M., Andersen Ø.E., Briscoe T. The BEA-2019 Shared Task on Grammatical Error Correction. BEA 2019. P. 52–75. DOI: 10.18653/v1/W19-4406.