Rationalizing Neural Predictions
Open accessThe DOI matches its Crossref record; 35 of the 35 references it cites were found in a registry or an index.
Abstract
Prediction without justification has limited applicability.As a remedy, we learn to extract pieces of input text as justifications -rationales -that are tailored to be short and coherent, yet sufficient for making the same prediction.Our approach combines two modular components, generator and encoder, which are trained to operate well together.The generator specifies a distribution over text fragments as candidate rationales and these are passed through the encoder for prediction.Rationales are never given during training.Instead, the model is regularized by desiderata for rationales.We evaluate the approach on multi-aspect sentiment analysis against manually annotated test cases.Our approach outperforms attention-based baseline by a significant margin.We also successfully illustrate the method on the question retrieval task. 1
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures






Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“A lot of work has also been done on argumentation and dialog games [6, 11, 51, 53] but the focus in these areas is generally the logical structure of the framework to express and to relate arguments or the protocol to exchange arguments.”
From Beyond explainability: justifiability and contestability of algorithmic decision systems (2022) · cites this paper as [53]
Experimental — accuracy not yet measuredIn the citing paper
“This model can achieve competitive accuracy while having built-in interpretability because the prediction is made exclusively based on rationales.”
From Human-AI Collaboration via Conditional Delegation: A Case Study of Content Moderation (2022) · cites this paper as [52]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
"Why Should I Trust You?": Explaining the Predictions of Any Classifier
Ribeiro, M. T., Singh, S., & Guestrin, C. (2016) · Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining · doi:10.1145/2939672.2939778 · Cited by 17,555
Neural Machine Translation by Jointly Learning to Align and Translate
Bahdanau, D., Cho, K., & Bengio, Y. (2014) · arXiv · doi:10.48550/arxiv.1409.0473 · Cited by 16,213 · cited 2×
“Attention based models offer another lens to the inner workings of neural models (Bahdanau et al., 2015; Cheng et al., 2016; Martins and Astudillo, 2016; Chen et al., 2015; Xu and Saenko, 2015; Yang et al., 2015).”
BackgroundConvolutional Neural Networks for Sentence Classification
Kim, Y. (2014) · Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.3115/v1/d14-1181 · Cited by 14,831
Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
Xu, K., Ba, J., Kiros, R., Cho, K., Courville, A., Salakhutdinov, R., Zemel, R., & Bengio, Y. (2015) · arXiv · doi:10.48550/arxiv.1502.03044 · Cited by 8,248
Bag of Tricks for Efficient Text Classification
Joulin, A., Grave, E., Bojanowski, P., & Mikolov, T. (2017) · Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers · doi:10.18653/v1/e17-2068 · Cited by 4,545
Bag of Tricks for Efficient Text Classification
Joulin, A., Grave, E., Bojanowski, P., & Mikolov, T. (2016) · arXiv · doi:10.48550/arxiv.1607.01759 · Cited by 4,542
A Convolutional Neural Network for Modelling Sentences
Kalchbrenner, N., Grefenstette, E., & Blunsom, P. (2014) · Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.3115/v1/p14-1062 · Cited by 3,601 · cited 1×
“For instance, we could use a convolutional network (Kim, 2014; Kalchbrenner et al., 2014), deep averaging network (Iyyer et al.”
MethodA Neural Attention Model for Abstractive Sentence Summarization
Rush, A. M., Chopra, S., & Weston, J. (2015) · Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d15-1044 · Cited by 2,656
Recurrent Models of Visual Attention
Mnih, V., Heess, N., Graves, A., & Kavukcuoglu, K. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1406.6247 · Cited by 2,568 · cited 1×
“space of possible rationales in terms of their interaction with the encoder. We could also apply variance reduction techniques to increase stability of stochastic training (cf. (Weaver and Tao, 2001; Mnih et al., 2014; Ba et al., 2015; Xu et al., 2015)). 7 Acknowledgments We thank Prof. Julian McAuley for sharing the…”
MethodTeaching Machines to Read and Comprehend
Hermann, K. M., Kočiský, T., Grefenstette, E., Espeholt, L., Kay, W., Suleyman, M., & Blunsom, P. (2015) · arXiv (Cornell University) · doi:10.48550/arxiv.1506.03340 · Cited by 2,521
Research works printed but not linked: 6
- [6]6. Cicero dos Santos, Luciano Barbosa, Dasha Bogdanova, and Bianca Zadrozny. 2015. Learning hybrid rep- resentations to retrieve semantically equivalent ques- tions. In Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Lan- guage Processing (Volume 2: Short Papers), pages 694-699, Beijing, China, July. Association for Com- putational Linguistics.
- [7]7. Manaal Faruqui, Jesse Dodge, Sujay K. Jauhar, Chris Dyer, Eduard Hovy, and Noah A. Smith. 2015a. Retrofitting word vectors to semantic lexicons. In Pro- ceedings of NAACL.
- [9]9. Aurélie Herbelot and Eva Maria Vecchi. 2015. Build- ing a shared world: mapping distributional to model- theoretic semantic spaces. In Proceedings of the 2015 Conference on Empirical Methods in Natural Lan- guage Processing. Association for Computational Lin- guistics.
- [12]12. Mohit Iyyer, Varun Manjunatha, Jordan Boyd-Graber, and Hal Daumé III. 2015. Deep unordered compo- sition rivals syntactic methods for text classification. In Proceedings of the 53rd Annual Meeting of the As- sociation for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers).
- [27]27. Tim Rocktäschel, Edward Grefenstette, Karl Moritz Her- mann, Tomáš Kočiskỳ, and Phil Blunsom. 2016. Rea- soning about entailment with neural attention. In In- ternational Conference on Learning Representations.
- [29]29. Sebastian Thrun. 1995. Extracting rules from artifi- cial neural networks with distributed representations. In Advances in neural information processing systems (NIPS).
Cited by
Items 501–600 of 900
Human Evaluation of Spoken vs. Visual Explanations for Open-Domain QA
Gonzalez, A. V., Bansal, G., Fan, A., Jia, R., Mehdad, Y., & Iyer, S. (2020) · arXiv · doi:10.48550/arxiv.2012.15075 · Cited by 7
“Lei et al. (2016) introduced an approach for extracting rationales by selecting phrases from the input text which are suf-ficient to provide an output.”
MethodExplanation User Interfaces: A Systematic Literature Review
Cappuccio, E., Esposito, A., Greco, F., Desolda, G., Lanzilotti, R., & Rinzivillo, S. (2026) · ACM Computing Surveys · doi:10.1145/3834765 · Cited by 9
“Furthermore, it explores the need for a human-centred process and its implications for XUIs.”
Does ChatGPT Spell the End of Automatic Question Generation Research?
Singh, M., Patvardhan, C., & Lakshmi, C. V. (2023) · IEEE International Conference on Computer Vision and Machine Intelligence (CVMI) · doi:10.1109/cvmi59935.2023.10464618 · Cited by 8
“Another reported work discusses techniques for generating rationales or explanations for predictions made by neural models, which can be relevant in the context of question generation [23].”
BackgroundTang, Z. S., & Surdeanu, M. (2023) · Computational Linguistics · doi:10.1162/coli_a_00463 · Cited by 12
“The fact that our method outperforms considerably the unsupervised rationale approach of Lei, Barzilay, and Jaakkola (2016), which is driven solely by relation classification performance, further emphasizes that a “human-in-the-loop” method such as ours is necessary to yield meaningful explanations.”
BackgroundMethodInfluentialDocument-Level Multi-Aspect Sentiment Classification for Online Reviews of Medical Experts
Shi, T., Rakesh, V., Wang, S., & Reddy, C. K. A. (2019) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3357384.3357828 · Cited by 7
“Although sophisticated models have been proposed for these tasks, they have only been applied to other types of datasets [15] and some of them have only been tested on small-scale datasets [14, 41].”
MethodRationalizing Medical Relation Prediction from Corpus-level Statistics
Wang, Z., Lee, J., Lin, S. M., & Sun, H. (2020) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2020.acl-main.719 · Cited by 7
Adversarial Separation Network for Text Style Transfer
Yang, H., Zhou, G., & He, T. (2021) · ACM Transactions on Asian and Low-Resource Language Information Processing · doi:10.1145/3472621 · Cited by 7
“According to the theory of VAE, we need to compute a lower bound—the Evidence Lower Bound (ELBO) [13] and alternatively maximize it to optimize our subjective.”
BackgroundMethodInfluentialLiu, B. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108639286.002 · Cited by 7
Argumentative explanations for pattern-based text classifiers
Lertvittayakumjorn, P., & Toni, F. (2023) · Argument & Computation · doi:10.3233/aac-220004 · Cited by 8
“Focusing on local explanation methods (aiming to explain specific predictions), we can see several forms of explanations in literature such as extracted rationales [50, 51], attribution scores [52, 53], rules [54, 55], influential training examples [56, 57], and counterfactual examples [58, 59].”
MethodContextualized medication event extraction with levitated markers
Vasilakes, J., Georgiadis, P., Nguyen, N. T. H., Miwa, M., & Ananiadou, S. (2023) · Journal of Biomedical Informatics · doi:10.1016/j.jbi.2023.104347 · Cited by 7
Qiao, C., Gao, B., Liu, Y., Hu, X., Hu, W., Calhoun, V. D., & Wang, Y. (2023) · Medical Image Analysis · doi:10.1016/j.media.2023.102941 · Cited by 7
Hsu, S. T., Moon, C., Jones, P., & Samatova, N. F. (2018) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v32i1.11972 · Cited by 7
Nutri-bullets: Summarizing Health Studies by Composing Segments
Shah, D., Yu, L., Lei, T., & Barzilay, R. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i15.17624 · Cited by 8
Self-citationFew-Shot Upsampling for Protest Size Detection
Halterman, A., & Radford, B. J. (2021) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2021.findings-acl.325 · Cited by 9
Cross-Domain Transfer of Generative Explanations Using Text-to-Text Models
Erliksson, K. F., Arpteg, A., Matskin, M., & Payberah, A. H. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-80599-9_8 · Cited by 8
“Another line of work for explainable NLP is based on rationalization pipelines that aim to produce extractive explanations by splitting the prediction process into two subsequent modules; a rationale extractor and a predictor [1,10,25,34].”
BackgroundRaFoLa: A Rationale-Annotated Corpus for Detecting Indicators of Forced Labour
Guzman, E. M., Schlegel, V., & Batista-Navarro, R. (2022) · Language Resources and Evaluation Conference (LREC) · doi:10.63317/4i2g7vc3hjbg · Cited by 12
Kott, A., Thomas, R. W., Drašar, M., Kont, M., Poylisher, A., Blakely, B., Théron, P., Evans, N., Leslie, N., Singh, R., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1804.07646 · Cited by 7
“For example, the DARPA Explainable Artificial Intelligence (XAI) program (Gunning 2016) focused on improving the ability of ML models to explain black box decision making to end-users providing the level of trust needed in operating with these autonomous systems (Lei et al. 2016).”
BackgroundInterpreting search result rankings through intent modelingPreprint
Singh, J., & Anand, A. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1809.05190 · Cited by 7
“Recently, in the machine learning (ML) and NLP communities there has been growing interest in the interpretability of ML models [3, 15, 17] but there has been limited work on interpreting retrieval models considered central to IR.”
BackgroundSelect and Attend: Towards Controllable Content Selection in Text GenerationPreprint
Shen, X., Suzuki, J., Inui, K., Su, H., Klakow, D., & Sekine, S. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.04453 · Cited by 8
“2Devlin et al. (2019) have shown that excellent performance can be obtained by assuming such conditionally independence given a sufficiently expressive representation of x, though modelling a richer inter-label dependency is for sure beneficial (Lei et al., 2016; Nallapati et al., 2017).”
BackgroundDuTrust: A Sentiment Analysis Dataset for Trustworthiness EvaluationPreprint
Wang, L., Liu, H., Peng, S., Tang, H., Xiao, X., Chen, Y., Wu, H. R., & Wang, H. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2108.13140 · Cited by 7
“Two-model methods use independent extraction and prediction modules, so as to first extract explanations from the input and then make predictions based on the extracted explanations (Lei et al. [2016], Jain et al. [2020]).”
MethodLearning Representations by Humans, for HumansPreprint
Hilgard, S., Rosenfeld, N., Banaji, M. R., Cao, J., & Parkes, D. C. (2019) · PsyArXiv (OSF Preprints) · doi:10.31234/osf.io/4nvts · Cited by 8
Unsupervised Selective Rationalization with Noise Injection
Storek, A., Subbiah, M., & McKeown, K. R. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.707 · Cited by 10
Canha, D., Kubler, S., Främling, K., & Fagherazzi, G. (2025) · ACM Computing Surveys · doi:10.1145/3737445 · Cited by 10
State Gradients for RNN Memory Analysis
Verwimp, L., hamme, H. V., Renkens, V., & Wambacq, P. (2018) · BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP (BlackboxNLP) · doi:10.18653/v1/w18-5443 · Cited by 6
“Lei et al. [17] extract the parts of the input that are important for predicting the output, while Alvarez-Melis and Jaakkola [18] provide a general framework that can explain the predictions of models with structured input and output.”
BackgroundExplainable AI and Its Applications in Healthcare
Sarkar, A. (2022) · Intelligent systems reference library · doi:10.1007/978-3-031-12807-3_6 · Cited by 6
Explainable YouTube Video Identification Using Sufficient Input Subsets
Afandi, W., Bukhari, S. M. A. H., Khan, M. U. S., Maqsood, T., Fayyaz, M. A. B., Ansari, A. R., & Nawaz, R. (2023) · IEEE Access · doi:10.1109/access.2023.3261562 · Cited by 6
“Although many explainable procedures, including [14] and [15] have demonstrated a model that is interpretable by humans.”
BackgroundZhang, H., Korikov, A., Farinneya, P., Pour, M. M. A., Bharadwaj, M., Pesaranghader, A., Huang, X. Y., Lok, Y. X., Wang, Z., Jones, N., et al. (2023) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3539618.3591880 · Cited by 15
“[19], they consider two-step “hard” extraction where an encoder identifies rationalizing spans and then an independent decoder uses these spans as inputs for predictions.”
BackgroundMethodLearning Robust Rationales for Model Explainability: A Guidance-Based Approach
Hu, S., & Yu, K. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i16.29783 · Cited by 6
“How-The ever, with increasing the skew threshold k , RNP fails to find human-annotated rationales, and its F1-score rapidly decreases in the palate aspect of the Skew-Predictor and Skew-Selector settings.”
BackgroundMethodInfluentialDeciphering Compatibility Relationships with Textual Descriptions via Extraction and Explanation
Wang, Y., He, Z., He, Z., Xu, H., & McAuley, J. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i8.28764 · Cited by 10
Resck, L., Raimundo, M. M., & Poco, J. (2024) · Findings of the Association for Computational Linguistics: NAACL · doi:10.18653/v1/2024.findings-naacl.262 · Cited by 7
“Noticeably, our approach also differs from previous work that rationalizes the input but does not leverage human annotations (Lei et al., 2016; Bastings et al., 2019; Jain et al., 2020).”
ResultExplainable Information Retrieval
Anand, A., Saha, S., & Venktesh, V. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-88720-8_40 · Cited by 6
On the Granularity of Explanations in Model Agnostic NLP Interpretability
Rychener, Y., Renard, X., Seddah, D., Frossard, P., & Detyniecki, M. (2023) · Communications in computer and information science · doi:10.1007/978-3-031-23618-1_33 · Cited by 6
“Interpretability of Natural Language Processing (NLP) models can be addressed by developing inherently interpretable classification models [14,5,10] or with Post-Hoc interpretability that can be applied to already trained models.”
MethodThe need for a numeric measure of explainability
Mulder, W. D., & Valcke, P. (2021) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata52589.2021.9671745 · Cited by 7
Kamath, U., & Liu, J. (2021) · Springer eBooks · doi:10.1007/978-3-030-83356-5_6 · Cited by 6
Game-Theoretic Interpretability for Temporal ModelingPreprint
Lee, G.-H., Alvarez-Melis, D., & Jaakkola, T. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1807.00130 · Cited by 6
“Our approach differs from interpretable models such as interpretation generators (Al-Shedivat et al., 2017; Lei et al., 2016) where the architecture or the function class is itself explicitly constrained towards interpretability, e.”
MethodSelf-citationHarnessing Explanations to Bridge AI and HumansPreprint
Lai, V., Carton, S., & Tan, C. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.07370 · Cited by 7
“(2016) [12].”
MethodSalKG: Learning From Knowledge Graph Explanations for Commonsense ReasoningPreprint
Chan, A., Xu, J., Long, B., Sanyal, S., Gupta, T., & Ren, X. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.08793 · Cited by 15
“…(Pruthi et al., 2020; Hase et al., 2020; Narang et al., 2020; Andreas et al., 2017), pruned inputs (Jain et al., 2020; Bastings et al., 2019; Lei et al., 2016), additional inputs (Hase and Bansal, 2021; Co-Reyes et al., 2018), and intermediate variables (Wiegreffe et al., 2020; Zhou et…”
BackgroundEvaluating Explanations for Reading Comprehension with Realistic Counterfactuals.Preprint
Ye, X., Nair, R., & Durrett, G. (2021) · arXiv (Cornell University) · Cited by 6
Nutri-bullets: Summarizing Health Studies by Composing SegmentsPreprint
Shah, D., Yu, L., Lei, T., & Barzilay, R. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.11921 · Cited by 6
“We model P ( Z | X ) as a span extraction task leveraging a rationale extraction system (Lei, Barzilay, and Jaakkola 2016).”
MethodSelf-citationPALRACE: Reading Comprehension Dataset with Human Data and Labeled RationalesPreprint
Zou, J., Zhang, Y., Jin, P., Luo, C., Pan, X., & Ding, N. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.12373 · Cited by 10
Laugwitz, L. L. (2021) · SocArXiv (OSF Preprints) · doi:10.31235/osf.io/gt28f · Cited by 6
Automatic Fact-Guided Sentence Modification
Shah, D., Schuster, T., & Barzilay, R. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i05.6406 · Cited by 36
Self-citationX-Troll: eXplainable Detection of State-Sponsored Information Operations Agents
Tian, L., Zhang, X., Kim, M., Biggs, J., & Rizoiu, M.-A. (2025) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3746252.3761028 · Cited by 7
Learnable Game-Theoretic Policy Optimization for Data-Centric Self-Explanation Rationalization
Zhao, Y., Wang, Z., Yu, X., Li, X., Liang, J., & Li, R. (2025) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2025.3638864 · Cited by 10
“Following previous research [15], [21], [34], we obtain BeerAdvocate [63], BeerAdvocate*[13], and HotelReview [64] datasets, which are all publicly available.”
InfluentialModular Chain-of-Thought (CoT) for LLM-Based Conceptual Construction Cost Estimation
Ghimire, P., Kim, K., Stentz, T. L., & Roy, T. (2026) · Buildings · doi:10.3390/buildings16020396 · Cited by 5
Variable Instance-Level Explainability for Text Classification.
Chrysostomou, G., & Αλέτρας, Ν. (2021) · White Rose Research Online (University of Leeds, The University of Sheffield, University of York) · Cited by 5
Gonçalves, D. M., Henriques, R., & Costa, R. S. (2021) · Cancers · doi:10.3390/cancers13133217 · Cited by 5
“Rationalizers propose a novel approach for incorporating rationale generation as an integral part of the overall learning process [91].”
BackgroundAspect-based Sentiment Analysis with Weighted Relational Graph Attention Network
Huo, Y., Jiang, D., & Hichem Sahli, B. (2021) · Companion Publication of the International Conference on Multimodal Interaction (ICMI Companion) · doi:10.1145/3461615.3491104 · Cited by 7
“Some of the recent efforts resorted to the study of attention mechanism [14, 31], which is employed to enhance the influence from the target-aspect to the final vector of the whole sentence for prediction.”
BackgroundTrigger-Argument based Explanation for Event Detection
Guan, Y., Chen, J., Lécué, F., Pan, J. Z., Li, J., & Li, R. (2023) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2023.findings-acl.312 · Cited by 5
“…there exist many works that focus on explaining the model behavior on natural language processing (NLP) problems, such as text classification (Lei et al., 2016), text matching (Jiang et al., 2021) and machine reading comprehension (Ju et al., 2021), very little progress has been made to…”
BackgroundOn Explanations for Hybrid Artificial Intelligence
Nolle, L., Stahl, F. T., & El-Mihoub, T. A. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-47994-6_1 · Cited by 6
All Things Considered: Detecting Partisan Events from News Media with Cross-Article Comparison
Liu, Y., Zhang, X., Zou, K., Huang, R., Beauchamp, N., & Wang, L. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.957 · Cited by 9
Explanation Guided Knowledge Distillation for Pre-trained Language Model Compression
Yang, Z., Zhang, Y., Sui, D., Ju, Y., Zhao, J., & Liu, K. (2024) · ACM Transactions on Asian and Low-Resource Language Information Processing · doi:10.1145/3639364 · Cited by 9
WELL: Applying bug detectors to bug localization via weakly supervised learning
Zhang, H., Li, Z., Li, J., Jin, Z. G., & Li, G. (2024) · Journal of Software Evolution and Process · doi:10.1002/smr.2669 · Cited by 5
“As for NLP, which is more relevant to our subject tasks, selective rationalization 37,38,39,40,41 is one of the most effective techniques to explain sentence classifiers.”
MethodFan, Y. B., Hu, Z., Li, R., Wang, Y. X., Zhang, G., Tan, H., & Liang, J. (2024) · Information Sciences · doi:10.1016/j.ins.2024.121763 · Cited by 7
Learning from shortcut: a shortcut-guided approach for explainable graph learning
Yue, L., Liu, Q., Liu, Y., Gao, W., & Yao, F. (2025) · Frontiers of Computer Science · doi:10.1007/s11704-024-40452-4 · Cited by 5
Rationale-based Human-in-the-Loop via Supervised Attention.
Kanchinadam, T., Westpfahl, K., You, Q., & Fung, G. (2020) · Knowledge Discovery and Data Mining · Cited by 11
Véstias, M. P. (2021) · IGI Global eBooks · doi:10.4018/978-1-6684-2408-7.ch077 · Cited by 8
Li, Y., Zhang, X., Qian, B., Gao, Z., Guan, C., Zheng, Y., Zheng, H., Wu, F., & Li, C. (2021) · Proceedings · doi:10.1109/icdm51629.2021.00045 · Cited by 7
When neuro-robots go wrong: A review
Khan, M. S., & Olds, J. L. (2023) · Frontiers in Neurorobotics · doi:10.3389/fnbot.2023.1112839 · Cited by 5
“Unsupervised (Lei et al., 2016; Esmaeili et al., 2019) Researchers assume no explicit annotations about input data.”
BackgroundComplex Reasoning in Natural Languag
Zhao, W., Geva, M., Lin, B. Y., Yasunaga, M., Madaan, A., & Yu, T. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-tutorials.2 · Cited by 11
Context-Aware Feature Selection and Classification
Wang, J., & Bilgic, M. (2023) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2023/480 · Cited by 6
Assistive Recipe Editing through Critiquing
Antognini, D., Li, S. X., Faltings, B. V., & McAuley, J. (2023) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2023.eacl-main.28 · Cited by 7
A Sequence-to-Structure Approach to Document-level Targeted Sentiment Analysis
Song, N., Cai, H., Xia, R., Yu, J., Wu, Z. Z., & Dai, X. (2023) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2023.findings-emnlp.515 · Cited by 5
“Lei et al. (2016) proposed an encoder-generator framework to extract rationales for aspect category and predict aspect category sentiment rating .”
BackgroundREFER: An End-to-end Rationale Extraction Framework for Explanation Regularization
Madani, M. R. G., & Minervini, P. (2023) · Conference on Computational Natural Language Learning · doi:10.18653/v1/2023.conll-1.40 · Cited by 11
Information bottleneck based knowledge selection for commonsense reasoning
Yang, Z., Zhang, Y., Cao, P., Cao, L., Chen, J., Zhao, J., & Liu, K. (2024) · Information Sciences · doi:10.1016/j.ins.2024.120134 · Cited by 7
Neural embeddings: accurate and readable inferences based on semantic kernels
Croce, D., Rossini, D., & Basili, R. (2019) · Natural Language Engineering · doi:10.1017/s1351324919000238 · Cited by 5
“…methodology exploits the coupling of the classifier with some sort of generator, or decoder, responsible for the selection of output justifications: Lei, Barzilay, and Jaakkola (2016) propose a generator that provides rationales for a multi-aspect sentiment analysis prediction by highlighting…”
MethodDesmond, M. J., Finegan‐Dollak, C., Boston, J., & Arnold, M. (2020) · Annual Meeting of the Association for Computational Linguistics: System Demonstrations (ACL Demo) · doi:10.18653/v1/2020.acl-demos.21 · Cited by 5
“Lei et al. (2016) jointly trained a generator and an encoder in order to generate rationales for sentiment prediction and a similar-question-retrieval task.”
MethodRANCC: Rationalizing Neural Networks via Concept Clustering
Bashier, H. K., Kim, M., & Goebel, R. G. (2020) · International Conference on Computational Linguistics · doi:10.18653/v1/2020.coling-main.286 · Cited by 6
“These methods focus on augmenting the neural network with another network architecture to uncover the rationale by finding out which portions of the input contribute most to the prediction of a target class (Lei et al., 2016; Bastings et al., 2019).”
BackgroundMethodLearning to Recover Reasoning Chains for Multi-Hop Question Answering via Cooperative Games
Feng, Y., Yu, M., Xiong, W., Guo, X., Huang, J., Chang, S., Campbell, M. S., Greenspan, M., & Zhu, X. (2021) · doi:10.21428/594757db.ce391ef3 · Cited by 7
Model-Agnostic Local Explanations with Genetic Algorithms for Text Classification
Du, Q. (2021) · Proceedings/Proceedings of the ... International Conference on Software Engineering and Knowledge Engineering · doi:10.18293/seke2021-040 · Cited by 5
Practical Benefits of Feature Feedback Under Distribution Shift
Katakkar, A., Yoo, C. H., Wang, W., Lipton, Z. C., & Kaushik, D. (2022) · BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP (BlackboxNLP) · doi:10.18653/v1/2022.blackboxnlp-1.29 · Cited by 5
A Study of Explainability Features to Scrutinize Faceted Filtering Results
Qu, J., Arguello, J., & Wang, Y. (2021) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3459637.3482409 · Cited by 9
“One approach is to visually highlight which parts of a document influenced the system to predict a specific category [25, 35].”
BackgroundMethodHuman-like Explanation for Text Classification With Limited Attention Supervision
Zhang, D., Şen, C., Thadajarassiri, J., Hartvigsen, T., Kong, X., & Rundensteiner, E. A. (2021) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata52589.2021.9671444 · Cited by 8
Internet of Medical Things (IoMT) & Secured Using Steganography for Development of Smart Society 5.0
Dhawan, S., Gupta, R., Rana, A. K., & Sharma, S. C. (2022) · Asset analytics · doi:10.1007/978-981-19-1689-2_11 · Cited by 6
Do Deep Neural Nets Display Human-like Attention in Short Answer Scoring?
Zeng, Z., Li, X., Gašević, D., & Chen, G. (2022) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2022.naacl-main.14 · Cited by 8
“In this strand of research, one common method is called test-based (Li et al., 2015; Jain and Wallace, 2019; Sun et al., 2020; Lei et al., 2016), which interprets a prediction by identifying relevant parts of input data that drive the prediction, e.g., words contained in a long sentence that play…”
MethodCognition-driven multimodal personality classification
Gao, X., Wang, J., Li, S., Zhang, M., & Zhou, G. (2022) · Science China Information Sciences · doi:10.1007/s11432-020-3307-3 · Cited by 5
“Specifically, Lei et al. [20] employed reinforcement learning to rationalize neural prediction for multi-aspect sentiment analysis.”
BackgroundExClaim: Explainable Neural Claim Verification Using Rationalization
Gurrapu, S., Huang, L., & Batarseh, F. A. (2022) · IEEE Annual Software Technology Conference (STC) · doi:10.1109/stc55697.2022.00012 · Cited by 10
“Paper [32] categorizes this as an explain-then-predict model and [33]–[35] cite improvements in task performance optimization after using similar techniques.”
MethodNews-Based Sparse Machine Learning Models for Adaptive Asset Pricing
Zhu, L., Wu, H., & Wells, M. T. (2023) · Data Science in Science · doi:10.1080/26941899.2023.2187895 · Cited by 5
Unsupervised Explanation Generation via Correct Instantiations
Cheng, S., Wu, Z., Chen, J., Li, Z., Liu, Y., & Kong, L. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v37i11.26494 · Cited by 5
Explainable Text Classification for Legal Document Review in Construction Delay Disputes
Huber-Fliflet, N., Zhang, J., Gronvall, P., Wei, F., Spinelli, P., Dąbrowski, A. G., & Yang, J. (2023) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata59044.2023.10386240 · Cited by 5
“[13] proposed an approach to generate rationales for text classifications using a deep learning algorithm.”
MethodDifferentiable Greedy NetworksPreprint
Powers, T., Fakoor, R., Shakeri, S., Sethy, A., Kainth, A., Mohamed, A.-R., & Sarikaya, R. (2018) · arXiv.org · doi:10.48550/arxiv.1810.12464 · Cited by 6
“Evidence retrieval is a key part of claim verification or question answering which aims to provide reasoning and a measure of interpretability (Lei et al., 2016).”
BackgroundGuiding Extractive Summarization with Question-Answering RewardsPreprint
Arumae, K., & Liu, F. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.02321 · Cited by 6
“Our proposed method is inspired by the work of Lei et al. (2016) who seek to identify rationales from textual input to support sentiment classification and question retrieval.”
MethodLearning by Self-Explanation, with Application to Neural Architecture SearchPreprint
Hosseini, R., & Xie, P. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.12899 · Cited by 5
Rationally Inattentive Utility Maximization for Interpretable Deep Image ClassificationPreprint
Pattanayak, K., & Krishnamurthy, V. (2021) · RePEc: Research Papers in Economics · doi:10.48550/arxiv.2102.04594 · Cited by 6
“Another approach seeks to provide local approximations to the trained model, local w.r.t the input image (Lei et al., 2016; Lundberg & Lee, 2017).”
BackgroundMaking a (Counterfactual) Difference One Rationale at a TimePreprint
Plyler, M., Green, M. A., & Chi, M. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2201.05177 · Cited by 5
“The rationale framework used in this work was introduced in [19] and connected to the MMI criteria by Chen et al.”
BackgroundMethodInfluentialAutoCAD: Automatically Generating Counterfactuals for Mitigating Shortcut LearningPreprint
Wen, J., Zhu, Y., Zhang, J., Zhou, J., & Huang, M. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2211.16202 · Cited by 5
“Some recent studies attempt to jointly optimize a rationale extractor and a classifier in an end-to-end fashion (Lei et al., 2016; Chang et al., 2020; Paranjape et al., 2020; Yu et al., 2021).”
BackgroundGive Me More Details: Improving Fact-Checking with Latent RetrievalPreprint
Hu, X., Chen, J., Guo, Z., & Yu, P. S. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2305.16128 · Cited by 9
“On the other hand, LP-SparseMAP is more stable when compared with other latent variable models (Lei et al., 2016; Bast-ings et al., 2019), which rely on sampling-based gradient estimators and thus exhibit high variance.”
BackgroundMethodAdversarial Cooperative Rationalization: The Risk of Spurious Correlations in Even Clean DatasetsPreprint
Liu, W., Niu, Z., Gao, L., Deng, Z., Wang, J., Wang, H., & Li, R. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2505.02118 · Cited by 10
“The basic cooperative framework of ratio-nalization named RNP (Lei et al., 2016) is flexible and offers a unique advantage: certification of exclusion, which means any unselected input is guaranteed to have no contribution to prediction, making it important to the NLP community (Yu et al., 2021).”
Quantifying Factual Divergence in Generative Models: SHAP-LIME Based Hallucination Score for LLMs
Haq, I., Saqib, M., Zhang, Y., & Khan, I. A. (2026) · Multimedia Systems · doi:10.1007/s00530-025-02150-4 · Cited by 4
“Initial progress in XAI for NLP focused on classification-based tasks such as sentiment analysis, toxicity detection, and rationalized prediction [20].”
Narrix: Remixing Narrative Strategies from Examples for Story Writing
Zhang, C., Guo, S., Davis, A., & Koh, E. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3790813 · Cited by 6
“…evidence, and hallucinations) and are supported by NLP fndings that extractive, source-attributed justi-fcations improve factual faithfulness and auditability [53], while verifcation-style prompting has been shown to reduce hallucinations by forcing consistency checks against evidence [21].”
IDEAL: A malicious traffic detection framework with explanation-guided learning
Jia, H., Lang, B., Li, X., & Yan, Y. (2025) · Knowledge-Based Systems · doi:10.1016/j.knosys.2025.113419 · Cited by 6
Rao, Y., & Yang, F. (2022) · Wireless Communications and Mobile Computing · doi:10.1155/2022/5467572 · Cited by 5
“In this paper, we model education policy texts according to the importance of the text and the title to obtain an optimal vector representation of the text, which can be used to effectively classify education policy texts [17].”
MethodRetractedThe Problem of Sentiment Analysis
Liu, B. (2015) · Cambridge University Press eBooks · doi:10.1017/cbo9781139084789.003 · Cited by 5
Games for Fairness and Interpretability
Chu, E., Gillani, N., & Makini, S. P. (2020) · International World Wide Web Conference (WWW) · doi:10.1145/3366424.3384374 · Cited by 6
“While there is also work worth mentioning on (1) generating human readable explanations in natural langauge [18, 28], (2) distilling neural networks into more interpretable models such as decision”
BackgroundZero-Shot Rationalization by Multi-Task Transfer Learning from Question Answering
Kung, P.-N., Yang, T.-H., Chen, Y.-C., Yin, S.-S., & Chen, Y.-N. (2020) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2020.findings-emnlp.198 · Cited by 4
“Lei et al. (2016) is the first work that attempted to extract rationales in order to justify the model’s answers, where a rationale generator extracts the context and a predictor generates the answer based on the extracted rationales.”
BackgroundMethodInfluentialSpanPredict: Extraction of Predictive Document Spans with Neural Attention
Subramanian, V., Engelhard, M., Berchuck, S. I., Chen, L., Henao, R., & Carin, L. (2021) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2021.naacl-main.413 · Cited by 4
“(Lei et al., 2016) tackle this by introducing two networks, a generator and an encoder, which, respectively, filter for important words before making a prediction.”
BackgroundDeep Learning Architecture for Processing Cyber-Physical Data in the Electric Grid
Calzada, D., Hossain‐McKenzie, S., & Mao, Z. (2022) · IEEE Power and Energy Conference at Illinois (PECI) · doi:10.1109/peci54197.2022.9744015 · Cited by 4
Controlling the Focus of Pretrained Language Generation Models
Ji, J., Kim, Y., Glass, J. R., & He, T. (2022) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2022.findings-acl.260 · Cited by 6
Metadata
- Authors
- Lei, T., Barzilay, R., & Jaakkola, T.
- Affiliations
- MIT Computer Science and Artificial Intelligence Laboratory (US)
- Venue
- Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing
- Published
- 2016
- Pages
- 107-117
- OpenAlex
- W2440159969
- Cited by
- 777
- References
- 27
- Citations by year
- 2026: 27 · 2025: 33 · 2024: 39 · 2023: 88 · 2022: 92 · 2021: 138 · 2020: 126 · 2019: 101 · 2018: 78 · 2017: 39 · 2016: 4
- FWCI
- 39.87
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Conference on Empirical Methods in Natural Language Processing (EMNLP) (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- DSpace@MIT (Massachusetts Institute of Technology) (submittedVersion, cc-by-nc-sa)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Physical Sciences › Computer Science › Artificial Intelligence › Sentiment Analysis and Opinion Mining
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Keywords
- explanation generation, question retrieval, model regularization
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.