Differences

This shows you the differences between two versions of the page.

--- nlp:question_answering [2022/10/12 01:07] – jmflanig
+++ nlp:question_answering [2025/05/13 19:46] (current) – [Overviews] jmflanig
@@ Line 6: / Line 6: @@
   * **[[https://arxiv.org/ftp/arxiv/papers/2001/2001.01582.pdf|Baradaran et al 2020 - A Survey on Machine Reading Comprehension Systems]]**
   * [[https://arxiv.org/pdf/2010.00389.pdf|Thayaparan et al 2020 - A Survey on Explainability in Machine Reading Comprehension]]
+  * **[[https://arxiv.org/pdf/2107.12708.pdf|Rogers et al 2022 - QA Dataset Explosion: A Taxonomy of NLP Resources for Question Answering and Reading Comprehension]]** [[https://dl.acm.org/doi/pdf/10.1145/3560260|ACM version (better)]]
 ===== Demos =====
   * [[https://demo.allennlp.org/reading-comprehension/transformer-qa|AllenNLP - RoBERTa QA Model Online Demo]]
 ===== Key Papers =====
+  * Early papers
+    * [[https://aclanthology.org/W00-0603.pdf|Riloff & Thelen 2000 - A Rule-based Question Answering System for Reading Comprehension Tests]] (Cited by the SQuAD 1.0 paper)
   * [[https://arxiv.org/pdf/1606.02858v2.pdf|Chen et al 2016 - A Thorough Examination of the CNN/Daily Mail Reading Comprehension Task]]
+  * [[https://arxiv.org/pdf/1606.05250.pdf|Rajpurkar et al 2016 - SQuAD: 100,000+ Questions for Machine Comprehension of Text]]
   * BiDAF model
+  * [[https://arxiv.org/pdf/1806.03822.pdf|Rajpurkar et al 2018 - Know What You Don't Know: Unanswerable Questions for SQuAD]] (SQuAD 2.0 paper)
   * [[https://aclanthology.org/2020.emnlp-main.550.pdf|Karpukhin et al 2020 - Dense Passage Retrieval for Open-Domain Question Answering]]
+  * [[https://arxiv.org/pdf/2404.06283|Basmov et al 2024 - LLMs’ Reading Comprehension Is Affected by Parametric Knowledge and Struggles with Hypothetical Statements]]
 ====== Topics ======
 ===== General QA Papers =====
   * [[https://arxiv.org/pdf/1601.01705.pdf|Andreas et al 2016 - Learning to Compose Neural Networks for Question Answering]]
+  * [[https://arxiv.org/pdf/2404.06283|Basmov et al 2024 - LLMs’ Reading Comprehension Is Affected by Parametric Knowledge and Struggles with Hypothetical Statements]]
 ===== Explanation And Implicit Reasoning Papers =====
@@ Line 25: / Line 32: @@
   * [[https://arxiv.org/pdf/2101.02235.pdf|Geva et al 2021 - Did Aristotle Use a Laptop? A Question Answering Benchmark with Implicit Reasoning Strategies]]
   * [[https://arxiv.org/pdf/2104.08661.pdf|Dalvi et al 2021 - Explaining Answers with Entailment Trees]]
+===== QA with Attribution =====
+  * [[https://arxiv.org/pdf/2212.08037|Bohnet et al 2022 - Attributed Question Answering: Evaluation and Modeling for Attributed Large Language Models]]
 ===== Robust Question Answering =====
-  * [[https://arxiv.org/pdf/2004.14648.pdf|Robust Question Answering Through Sub-part Alignment]]
+  * [[https://arxiv.org/pdf/2004.14648.pdf|Chen & Durrett 2020 - Robust Question Answering Through Sub-part Alignment]]
 ===== Open-Domain Question Answering =====
@@ Line 45: / Line 56: @@
 ===== Yes/No Questions =====
   * [[https://arxiv.org/pdf/1905.10044.pdf|Clark et al 2019 - BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questions]]
+  * [[https://aclanthology.org/2022.naacl-main.79.pdf|Sulem et al 2021 - Yes, No or IDK: The Challenge of Unanswerable Yes/No Questions]]
 ===== Long-Form QA =====
@@ Line 64: / Line 76: @@
 ===== Domain Adaptation =====
-See the related work in [[https://arxiv.org/pdf/2203.08926.pdf|Yue 2022]], and also [[https://scholar.google.com/citations?user=VKKAfwMAAAAJ&hl=en|Arafat Sultan's publications on QA]].
+See the related work in [[https://arxiv.org/pdf/2203.08926.pdf|Yue 2022]] and [[https://arxiv.org/pdf/2210.10861.pdf|Yue 2022]], and also [[https://scholar.google.com/citations?user=VKKAfwMAAAAJ&hl=en|Arafat Sultan's publications on QA]].
 === Synthetic Question Generation ===
@@ Line 70: / Line 82: @@
   * [[https://aclanthology.org/D17-1090.pdf|Duan et al 2017 - Question Generation for Question Answering]]
-  * [[https://dl.acm.org/doi/abs/10.1145/3404835.3463085|2021 - Synthetic Target Domain Supervision for Open Retrieval QA]]
+  * [[https://arxiv.org/pdf/1706.02027.pdf|Tang et al 2017 - Question Answering and Question Generation as Dual Tasks]]
   * [[https://aclanthology.org/2020.emnlp-main.439.pdf|Shakeri et al 2020 - End-to-End Synthetic Data Generation for Domain Adaptation of Question Answering Systems]] Generates QA pairs in the target domain using encoder-decoder large pre-trained model fine-tuned to the dataset.
   * [[https://arxiv.org/pdf/2012.01414.pdf|Reddy et al 2020 - End-to-End QA on COVID-19: Domain Adaptation with Synthetic Training]]
+  * [[https://arxiv.org/pdf/2010.06028.pdf|Shakeri et al 2020 - End-to-End Synthetic Data Generation for Domain Adaptation of Question Answering Systems]]
   * **[[https://arxiv.org/pdf/2010.12776.pdf|Chen et al 2020 - Improved Synthetic Training for Reading Comprehension]]**
   * [[https://arxiv.org/pdf/2010.16021.pdf|2020 - CliniQG4QA: Generating Diverse Questions for Domain Adaptation of Clinical Question Answering]]
@@ Line 78: / Line 91: @@
   * [[https://arxiv.org/pdf/2203.08926.pdf|Yue et al 2022 - Synthetic Question Value Estimation
 for Domain Adaptation of Question Answering]]
+  * [[https://arxiv.org/pdf/2204.09248.pdf|Reddy et al 2022 - Synthetic Target Domain Supervision for Open Retrieval QA]]
+  * [[https://arxiv.org/pdf/2210.10861.pdf|Yue et al 2022 - QA Domain Adaptation using Hidden Space Augmentation and Self-Supervised Contrastive Adaptation]]
 ===== Cross-Lingual =====
@@ Line 89: / Line 104: @@
 ====== Datasets ======
-See also [[http://nlpprogress.com/english/question_answering.html|NLP Progress - Question Answering]] and [[https://docs.google.com/spreadsheets/d/1gWDy7-rfT0efhmFF42fR9cPpfqZDKl07q1JxrRqGEVI/edit#gid=0|Geetanjali's QA Datasets Spreadsheet]]
+See [[https://arxiv.org/pdf/2107.12708.pdf|Rogers et al 2022 - QA Dataset Explosion]]. See also [[http://nlpprogress.com/english/question_answering.html|NLP Progress - Question Answering]] and [[https://docs.google.com/spreadsheets/d/1gWDy7-rfT0efhmFF42fR9cPpfqZDKl07q1JxrRqGEVI/edit#gid=0|Geetanjali's QA Datasets Spreadsheet]]
   * **CNN/Daily Mail Reading Comprehension**
     * Large-scale cloze-style QA dataset constructed from news articles and their summaries
@@ Line 131: / Line 146: @@
   * Product Domain (Product-related Question Answering - PQA)
     * **Amazon-PQA** and **AmazonPQSim**: [[https://aclanthology.org/2021.naacl-main.23.pdf|Rozen et al 2021 - Answering Product-Questions by Utilizing Questions from Other Contextually Similar Products]] Worked with Yes/No questions only.  Datasets available [[https://registry.opendata.aws/|here]] (search for [[https://registry.opendata.aws/amazon-pqa/|Amazon-PQA]] or AmazonPQSim). Dataset incluses free-form, not extractive answers and yes/no questions.
+  * Research Paper Domain
+    * **Qasper**: [[https://arxiv.org/pdf/2105.03011.pdf|Dasigi et al 2021 - A Dataset of Information-Seeking Questions and Answers Anchored in Research Papers]] This dataset is really hard. Contains everything: abstractive, multispan extractive, yes / no, unanswerable questions, all from a very long context.
 ====== Resources ======