2000 character limit reached
Unsupervised multiple choices question answering via universal corpus (2402.17333v1)
Published 27 Feb 2024 in cs.CL
Abstract: Unsupervised question answering is a promising yet challenging task, which alleviates the burden of building large-scale annotated data in a new domain. It motivates us to study the unsupervised multiple-choice question answering (MCQA) problem. In this paper, we propose a novel framework designed to generate synthetic MCQA data barely based on contexts from the universal domain without relying on any form of manual annotation. Possible answers are extracted and used to produce related questions, then we leverage both named entities (NE) and knowledge graphs to discover plausible distractors to form complete synthetic samples. Experiments on multiple MCQA datasets demonstrate the effectiveness of our method.
- “BAG: Bi-directional attention entity graph convolutional network for multi-hop reasoning question answering,” in EMNLP, 2019.
- “DAGN: Discourse-aware graph network for logical reasoning,” in NAACL, 2021.
- “A survey for efficient open domain question answering,” in ACL, 2023.
- “Context modeling with evidence filter for multiple choice question answering,” in ICASSP, 2022.
- “Narrow down before selection: A dynamic exclusion model for multiple-choice qa,” in ICASSP, 2023.
- “Joint reasoning with knowledge subgraphs for multiple choice question answering,” Information Processing & Management, vol. 60, no. 3, pp. 103297, 2023.
- “Large language models are neurosymbolic reasoners,” in AAAI, 2024.
- “Template-based question generation from retrieved sentences for improved unsupervised question answering,” in ACL, 2020.
- “Unsupervised question answering by cloze translation,” in ACL, 2019.
- “Learning how to active learn: A deep reinforcement learning approach,” in EMNLP, 2017.
- “Harvesting and refining question-answer pairs for unsupervised QA,” in ACL, 2020.
- “Unsupervised multiple choices question answering: Start learning from basic knowledge,” in ACL, 2021.
- “Knowledge-driven distractor generation for cloze-style multiple choice questions,” in AAAI, 2021.
- “Roberta: A robustly optimized bert pretraining approach,” arXiv preprint arXiv:1907.11692, 2019.
- “Better distractions: Transformer-based distractor generation and multiple choice question filtering,” arXiv preprint arXiv:2010.09598, 2020.
- “Self-supervised knowledge triplet learning for zero-shot question answering,” in EMNLP, 2020.
- “spacy: Industrial-strength natural language processing in python,” 2020.
- “Zero-shot commonsense question answering with cloze translation and consistency optimization,” arXiv preprint arXiv:2201.00136, 2022.
- “Phrase-based & neural unsupervised machine translation,” in EMNLP, 2018.
- “Multiple choice question corpus analysis for distractor characterization,” in LREC, 2014.
- “QA-GNN: Reasoning with language models and knowledge graphs for question answering,” in NAACL-HLT, 2021.
- “Conceptnet 5.5: an open multilingual graph of general knowledge,” in AAAI, 2017.
- “Scalable multi-hop relational reasoning for knowledge-aware question answering,” in EMNLP, 2020.
- “Adam: A method for stochastic optimization,” in ICLR, 2015.
- “Llama 2: Open foundation and fine-tuned chat models,” arXiv preprint arXiv:2307.09288, 2023.
- “Lora: Low-rank adaptation of large language models,” arXiv preprint arXiv:2106.09685, 2021.
- “SWAG: A large-scale adversarial dataset for grounded commonsense inference,” in EMNLP, 2018.
- “Think you have solved question answering? try arc, the ai2 reasoning challenge,” ArXiv, vol. abs/1803.05457, 2018.
- “CommonsenseQA: A question answering challenge targeting commonsense knowledge,” in NAACL-HLT, 2019.
- “Social IQa: Commonsense reasoning about social interactions,” in EMNLP-IJCNLP, 2019.
- “MCTest: A challenge dataset for the open-domain machine comprehension of text,” in EMNLP, 2013.
- “Llama: Open and efficient foundation language models,” arXiv preprint arXiv:2302.13971, 2023.
- Qin Zhang (98 papers)
- Hao Ge (49 papers)
- Xiaojun Chen (100 papers)
- Meng Fang (100 papers)