{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,16]],"date-time":"2026-01-16T11:25:14Z","timestamp":1768562714431,"version":"3.49.0"},"reference-count":62,"publisher":"MIT Press - Journals","license":[{"start":{"date-parts":[[2022,3,25]],"date-time":"2022-03-25T00:00:00Z","timestamp":1648166400000},"content-version":"vor","delay-in-days":83,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["direct.mit.edu"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2022,3,18]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>We introduce a neuro-symbolic natural logic framework based on reinforcement learning with introspective revision. The model samples and rewards specific reasoning paths through policy gradient, in which the introspective revision algorithm modifies intermediate symbolic reasoning steps to discover reward-earning operations as well as leverages external knowledge to alleviate spurious reasoning and training inefficiency. The framework is supported by properly designed local relation models to avoid input entangling, which helps ensure the interpretability of the proof paths. The proposed model has built-in interpretability and shows superior capability in monotonicity inference, systematic generalization, and interpretability, compared with previous models on the existing datasets.<\/jats:p>","DOI":"10.1162\/tacl_a_00458","type":"journal-article","created":{"date-parts":[[2022,3,21]],"date-time":"2022-03-21T19:09:55Z","timestamp":1647889795000},"page":"240-256","update-policy":"https:\/\/doi.org\/10.1162\/mitpressjournals.corrections.policy","source":"Crossref","is-referenced-by-count":3,"title":["Neuro-symbolic Natural Logic with Introspective Revision for Natural\n                    Language Inference"],"prefix":"10.1162","volume":"10","author":[{"given":"Yufei","family":"Feng","sequence":"first","affiliation":[{"name":"Ingenuity Labs Research Institute & ECE, Queen\u2019s University, Canada. feng.yufei@queensu.ca"}]},{"given":"Xiaoyu","family":"Yang","sequence":"additional","affiliation":[{"name":"Ingenuity Labs Research Institute & ECE, Queen\u2019s University, Canada. xiaoyu.yang@queensu.ca"}]},{"given":"Xiaodan","family":"Zhu","sequence":"additional","affiliation":[{"name":"Ingenuity Labs Research Institute & ECE, Queen\u2019s University, Canada. xiaodan.zhu@queensu.ca"}]},{"given":"Michael","family":"Greenspan","sequence":"additional","affiliation":[{"name":"Ingenuity Labs Research Institute & ECE, Queen\u2019s University, Canada. michael.greenspan@queensu.ca"}]}],"member":"281","published-online":{"date-parts":[[2022,3,18]]},"reference":[{"key":"2022033118512416700_bib1","first-page":"166","article-title":"Modular multitask reinforcement learning\n                        with policy sketches","volume-title":"International Conference on\n                        Machine Learning","author":"Andreas","year":"2017"},{"key":"2022033118512416700_bib2","article-title":"Hindsight experience\n                        replay","volume-title":"Advances in Neural Information Processing\n                        Systems","author":"Andrychowicz","year":"2017"},{"key":"2022033118512416700_bib3","doi-asserted-by":"publisher","DOI":"10.3115\/v1\/D14-1059","article-title":"Naturalli: Natural logic inference for\n                        common sense reasoning","volume-title":"Proceedings of the 2014\n                        Conference on Empirical Methods in Natural Language Processing\n                        (EMNLP)","author":"Angeli","year":"2014"},{"issue":"1","key":"2022033118512416700_bib4","doi-asserted-by":"publisher","first-page":"57","DOI":"10.1023\/A:1008203301671","article-title":"Language of thought: The connectionist\n                        contribution","volume":"7","author":"Aydede","year":"1997","journal-title":"Minds and Machines"},{"key":"2022033118512416700_bib5","doi-asserted-by":"publisher","first-page":"37","DOI":"10.1075\/llsee.25.06ben","article-title":"The semantics of variety in categorial\n                        grammar","volume":"25","author":"van Benthem","year":"1988","journal-title":"Categorial Grammar"},{"key":"2022033118512416700_bib6","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D15-1075","article-title":"A large annotated corpus for learning\n                        natural language inference","volume-title":"Proceedings of the\n                        2015 Conference on Empirical Methods in Natural Language Processing\n                        (EMNLP)","author":"Bowman","year":"2015"},{"key":"2022033118512416700_bib7","article-title":"e-SNLI: Natural language inference with\n                        natural language explanations","volume-title":"Advances in Neural\n                        Information Processing Systems","author":"Camburu","year":"2018"},{"key":"2022033118512416700_bib8","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P18-1224","article-title":"Neural natural language inference models\n                        enhanced with external knowledge","volume-title":"Proceedings of\n                        the 56th Annual Meeting of the Association for Computational Linguistics\n                        (ACL)","author":"Chen","year":"2017"},{"key":"2022033118512416700_bib9","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P17-1152","article-title":"Enhanced lstm for natural language\n                        inference","volume-title":"Proceedings of the 55th Annual Meeting\n                        of the Association for Computational Linguistics (ACL)","author":"Chen","year":"2017"},{"key":"2022033118512416700_bib10","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.starsem-1.7","article-title":"Neurallog: Natural language inference with joint neural and\n                        logical reasoning","author":"Chen","year":"2021","journal-title":"arXiv preprint\n                        arXiv:2105.14167"},{"key":"2022033118512416700_bib11","doi-asserted-by":"publisher","DOI":"10.1007\/11736790_9","article-title":"The PASCAL recognising textual entailment\n                        challenge","volume-title":"Proceedings of the First International\n                        Conference on Machine Learning Challenges: Evaluating Predictive Uncertainty\n                        Visual Object Classification, and Recognizing Textual Entailment","author":"Dagan","year":"2005"},{"key":"2022033118512416700_bib12","article-title":"Neuro-symbolic= neural+ logical+\n                        probabilistic","volume-title":"NeSy\u201919@ IJCAI, the 14th\n                        International Workshop on Neural-Symbolic Learning and Reasoning","author":"Raedt","year":"2019"},{"key":"2022033118512416700_bib13","article-title":"BERT: Pre-training of deep bidirectional\n                        transformers for language understanding","volume-title":"Proceedings of the 2019 Conference of the North American Chapter of\n                        the Association for Computational Linguistics: Human Language Technologies\n                        (NAACL-HLT)","author":"Devlin","year":"2019"},{"key":"2022033118512416700_bib14","doi-asserted-by":"publisher","first-page":"4443","DOI":"10.18653\/v1\/2020.acl-main.408","article-title":"Eraser: A benchmark to evaluate\n                        rationalized NLP models","volume-title":"Proceedings of the 58th\n                        Annual Meeting of the Association for Computational Linguistics","author":"DeYoung","year":"2020"},{"key":"2022033118512416700_bib15","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1613\/jair.5714","article-title":"Learning explanatory rules from noisy\n                        data","volume":"61","author":"Evans","year":"2018"},{"key":"2022033118512416700_bib16","doi-asserted-by":"publisher","first-page":"1172","DOI":"10.18653\/v1\/2020.coling-main.101","article-title":"Exploring end-to-end differentiable natural logic\n                        modeling","volume-title":"Proceedings of the 28th International\n                        Conference on Computational Linguistics","author":"Feng","year":"2020"},{"issue":"1-2","key":"2022033118512416700_bib17","doi-asserted-by":"publisher","first-page":"3","DOI":"10.1016\/0010-0277(88)90031-5","article-title":"Connectionism and cognitive architecture:\n                        A critical analysis","volume":"28","author":"Fodor","year":"1988","journal-title":"Cognition"},{"key":"2022033118512416700_bib18","article-title":"Neural-symbolic learning and reasoning:\n                        contributions and challenges","volume-title":"2015 AAAI Spring\n                        Symposium Series","author":"Garcez","year":"2015"},{"key":"2022033118512416700_bib19","doi-asserted-by":"publisher","first-page":"4485","DOI":"10.18653\/v1\/D19-1456","article-title":"Posing fair generalization tasks for natural\n                        language inference","volume-title":"Proceedings of the 2019\n                        Conference on Empirical Methods in Natural Language Processing and the 9th\n                        International Joint Conference on Natural Language Processing\n                        (EMNLP-IJCNLP)","author":"Geiger","year":"2019"},{"key":"2022033118512416700_bib20","doi-asserted-by":"publisher","first-page":"163","DOI":"10.18653\/v1\/2020.blackboxnlp-1.16","article-title":"Neural natural language inference models\n                        partially embed theories of lexical entailment and negation","volume-title":"Proceedings of the Third BlackboxNLP Workshop on Analyzing and\n                        Interpreting Neural Networks for NLP","author":"Geiger","year":"2020"},{"key":"2022033118512416700_bib21","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.acl-main.177","article-title":"Probing linguistic\n                        systematicity","volume-title":"ACL 2020","author":"Goodwin","year":"2020"},{"key":"2022033118512416700_bib22","doi-asserted-by":"publisher","first-page":"2385","DOI":"10.24963\/ijcai.2019\/331","article-title":"Using natural language for reward shaping\n                        in reinforcement learning","volume-title":"Proceedings of the\n                        Twenty-Eighth International Joint Conference on Artificial Intelligence,\n                        IJCAI-19","author":"Goyal","year":"2019"},{"key":"2022033118512416700_bib23","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/N18-2017","article-title":"Annotation artifacts in natural language\n                        inference data","volume-title":"NAACL-HLT(2)","author":"Gururangan","year":"2018"},{"key":"2022033118512416700_bib24","doi-asserted-by":"publisher","DOI":"10.5281\/zenodo.1212303","article-title":"spaCy: Industrial-strength Natural Language Processing in\n                        Python","author":"Honnibal","year":"2020"},{"key":"2022033118512416700_bib25","article-title":"MonaLog: A lightweight system for natural\n                        language inference based on monotonicity","volume-title":"Proceedings of the Society for Computation in Linguistics\n                        2020","author":"Hai","year":"2020"},{"key":"2022033118512416700_bib26","doi-asserted-by":"publisher","DOI":"10.1007\/s11225-012-9425-8","article-title":"Inclusion and exclusion in natural\n                        language","author":"Icard","year":"2012","journal-title":"Studia Logica"},{"key":"2022033118512416700_bib27","doi-asserted-by":"crossref","DOI":"10.33011\/lilt.v9i.1325","article-title":"Recent progress on monotonicity","volume-title":"Linguistic Issues in Language Technology","author":"Icard","year":"2014"},{"key":"2022033118512416700_bib28","doi-asserted-by":"publisher","first-page":"4198","DOI":"10.18653\/v1\/2020.acl-main.386","article-title":"Towards faithfully interpretable NLP\n                        systems: How should we define and evaluate faithfulness?","volume-title":"Proceedings of the 58th Annual Meeting of the Association for\n                        Computational Linguistics","author":"Jacovi","year":"2020"},{"key":"2022033118512416700_bib29","doi-asserted-by":"publisher","first-page":"5235","DOI":"10.18653\/v1\/2020.coling-main.459","article-title":"Hy-NLI: A hybrid system for natural\n                        language inference","volume-title":"Proceedings of the 28th\n                        International Conference on Computational Linguistics","author":"Kalouli","year":"2020"},{"key":"2022033118512416700_bib30","first-page":"2873","article-title":"Generalization without systematicity: On\n                        the compositional skills of sequence-to-sequence recurrent\n                        networks","volume-title":"International conference on machine\n                        learning","author":"Lake","year":"2018"},{"key":"2022033118512416700_bib31","doi-asserted-by":"publisher","DOI":"10.1007\/BF00413602","article-title":"Linguistics and natural\n                        logic","author":"Lakoff","year":"1970","journal-title":"Synthese"},{"key":"2022033118512416700_bib32","doi-asserted-by":"publisher","first-page":"107","DOI":"10.18653\/v1\/D16-1011","article-title":"Rationalizing neural\n                        predictions","volume-title":"Proceedings of the 2016 Conference\n                        on Empirical Methods in Natural Language Processing","author":"Lei","year":"2016"},{"key":"2022033118512416700_bib33","first-page":"5884","article-title":"Closed loop neural-symbolic learning via integrating neural\n                        perception, grammar parsing, and symbolic reasoning","volume-title":"International Conference on Machine Learning","author":"Li","year":"2020"},{"key":"2022033118512416700_bib34","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P17-1003","article-title":"Neural symbolic machines: Learning\n                        semantic parsers on Freebase with weak supervision","volume-title":"Proceedings of the 55th Annual Meeting of the Association for\n                        Computational Linguistics (Volume 1: Long Papers)","author":"Liang","year":"2017"},{"key":"2022033118512416700_bib35","unstructured":"Bill\n              MacCartney\n            \n          .\n                        2009. Natural Language\n                    Inference. Ph.D. thesis,\n                        Stanford University."},{"key":"2022033118512416700_bib36","doi-asserted-by":"publisher","first-page":"802","DOI":"10.3115\/1613715.1613817","article-title":"A phrase-based alignment model for natural\n                        language inference","volume-title":"Proceedings of the 2008\n                        Conference on Empirical Methods in Natural Language Processing","author":"MacCartney","year":"2008"},{"key":"2022033118512416700_bib37","doi-asserted-by":"publisher","first-page":"140","DOI":"10.3115\/1693756.1693772","article-title":"An extended model of natural\n                        logic","volume-title":"Proceedings of the Eight International\n                        Conference on Computational Semantics","author":"MacCartney","year":"2009"},{"key":"2022033118512416700_bib38","article-title":"The neuro-symbolic concept learner:\n                        Interpreting scenes, words, and sentences from natural\n                        supervision","volume-title":"International Conference on Learning\n                        Representations","author":"Mao","year":"2018"},{"key":"2022033118512416700_bib39","doi-asserted-by":"publisher","first-page":"4942","DOI":"10.1109\/CVPR.2018.00519","article-title":"Transparency by design: Closing the gap\n                        between performance and interpretability in visual\n                    reasoning","volume-title":"Proceedings of the IEEE Conference on\n                        Computer Vision and Pattern Recognition","author":"Mascharka","year":"2018"},{"key":"2022033118512416700_bib40","volume-title":"WordNet: An electronic lexical database","author":"Miller","year":"1998"},{"key":"2022033118512416700_bib41","article-title":"Computing relative polarity for textual\n                        inference","volume-title":"Proceedings of the Fifth International\n                        Workshop on Inference in Computational Semantics\n                    (icos-5)","author":"Nairn","year":"2006"},{"key":"2022033118512416700_bib42","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P19-1467","article-title":"Neural network alignment for sentential\n                        paraphrases","volume-title":"Proceedings of the 57th Annual\n                        Meeting of the Association for Computational Linguistics","author":"Ouyang","year":"2019"},{"key":"2022033118512416700_bib43","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/S18-2023","article-title":"Hypothesis only baselines in natural language\n                        inference","author":"Poliak","year":"2018","journal-title":"arXiv preprint arXiv:1805.01042"},{"key":"2022033118512416700_bib44","article-title":"Data-efficient deep reinforcement learning\n                        for dexterous manipulation","author":"Popov","year":"2017","journal-title":"arXiv preprint\n                        arXiv:1704.03073"},{"key":"2022033118512416700_bib45","article-title":"Improving language understanding by\n                        generative pre-training","author":"Radford","year":"2018"},{"issue":"8","key":"2022033118512416700_bib46","first-page":"9","article-title":"Language models are unsupervised multitask\n                        learners","volume":"1","author":"Radford","year":"2019","journal-title":"OpenAI blog"},{"key":"2022033118512416700_bib47","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v34i05.6397","article-title":"Probing natural language inference models\n                        through semantic fragments","volume-title":"Proceedings of the\n                        Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI)","author":"Richardson","year":"2020"},{"key":"2022033118512416700_bib48","first-page":"4344","article-title":"Learning by playing solving sparse reward\n                        tasks from scratch","volume-title":"International Conference on\n                        Machine Learning","author":"Riedmiller","year":"2018"},{"key":"2022033118512416700_bib49","article-title":"End-to-end differentiable\n                        proving","volume-title":"Proceedings of the 31st International\n                        Conference on Neural Information Processing Systems (NeurIPS)","author":"Rockt\u00e4schel","year":"2017"},{"key":"2022033118512416700_bib50","article-title":"Keeping your distance: Solving sparse\n                        reward tasks using self-balancing shaped rewards","volume-title":"NeurIPS","author":"Trott","year":"2019"},{"key":"2022033118512416700_bib51","volume-title":"Studies on natural logic and categorial\n                        grammar","author":"S\u00e1nchez Valencia","year":"1991"},{"key":"2022033118512416700_bib52","doi-asserted-by":"publisher","DOI":"10.1007\/978-94-009-4540-1","volume-title":"Essays in logical semantics","author":"Benthem","year":"1986"},{"key":"2022033118512416700_bib53","volume-title":"Language in Action: categories, lambdas and\n                        dynamic logic","author":"Benthem","year":"1995"},{"key":"2022033118512416700_bib54","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P19-1618","article-title":"Nlprolog: Reasoning with weak unification\n                        for question answering in natural language","volume-title":"Proceedings of the 57th Annual Meeting of the Association for\n                        Computational Linguistics (ACL)","author":"Weber","year":"2019"},{"key":"2022033118512416700_bib55","doi-asserted-by":"publisher","first-page":"1112","DOI":"10.18653\/v1\/N18-1101","article-title":"A broad-coverage challenge corpus for\n                        sentence understanding through inference","volume-title":"Proceedings of the 2018 Conference of the North American Chapter of\n                        the Association for Computational Linguistics: Human Language Technologies,\n                        Volume 1 (Long Papers)","author":"Williams","year":"2018"},{"issue":"3","key":"2022033118512416700_bib56","doi-asserted-by":"publisher","first-page":"229","DOI":"10.1023\/A:1022672621406","article-title":"Simple statistical gradient-following\n                        algorithms for connectionist reinforcement learning","volume":"8","author":"Williams","year":"1992","journal-title":"Machine Learning"},{"key":"2022033118512416700_bib57","doi-asserted-by":"publisher","first-page":"6105","DOI":"10.18653\/v1\/2020.acl-main.543","article-title":"Do neural models learn systematicity of monotonicity\n                        inference in natural language?","volume-title":"Proceedings of the\n                        58th Annual Meeting of the Association for Computational\n                        Linguistics","author":"Yanaka","year":"2020"},{"key":"2022033118512416700_bib58","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/W19-4804","article-title":"Can neural networks understand monotonicity\n                        reasoning?","volume-title":"Proceedings of the 2019 ACL Workshop\n                        BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP","author":"Yanaka","year":"2019"},{"key":"2022033118512416700_bib59","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/S19-1027","article-title":"Help: A dataset for identifying shortcomings of neural models\n                        in monotonicity reasoning","volume-title":"Proceedings of the\n                        Eighth Joint Conference on Lexical and Computational Semantics\n                        (*SEM)","author":"Yanaka","year":"2019"},{"key":"2022033118512416700_bib60","article-title":"Differentiable learning of logical rules for\n                        knowledge base reasoning","volume-title":"Advances in Neural\n                        Information Processing Systems","author":"Yang","year":"2017"},{"key":"2022033118512416700_bib61","first-page":"702","article-title":"A lightweight and high performance monolingual\n                        word aligner","volume-title":"Proceedings of the 51st Annual\n                        Meeting of the Association for Computational Linguistics (Volume 2: Short\n                        Papers)","author":"Yao","year":"2013"},{"key":"2022033118512416700_bib62","first-page":"1039","article-title":"Neural-symbolic VQA: Disentangling\n                        reasoning from vision and language understanding","volume-title":"Advances in Neural Information Processing Systems","author":"Yi","year":"2018"}],"container-title":["Transactions of the Association for Computational Linguistics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/direct.mit.edu\/tacl\/article-pdf\/doi\/10.1162\/tacl_a_00458\/2004304\/tacl_a_00458.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/direct.mit.edu\/tacl\/article-pdf\/doi\/10.1162\/tacl_a_00458\/2004304\/tacl_a_00458.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,3,31]],"date-time":"2022-03-31T23:53:07Z","timestamp":1648770787000},"score":1,"resource":{"primary":{"URL":"https:\/\/direct.mit.edu\/tacl\/article\/doi\/10.1162\/tacl_a_00458\/110011\/Neuro-symbolic-Natural-Logic-with-Introspective"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022]]},"references-count":62,"URL":"https:\/\/doi.org\/10.1162\/tacl_a_00458","relation":{},"ISSN":["2307-387X"],"issn-type":[{"value":"2307-387X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2022]]},"published":{"date-parts":[[2022]]}}}