{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,30]],"date-time":"2026-06-30T14:29:51Z","timestamp":1782829791500,"version":"3.54.5"},"reference-count":56,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,4,8]],"date-time":"2026-04-08T00:00:00Z","timestamp":1775606400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Artif. Intell."],"abstract":"\n Analyzing the tone of Bangla financial news is challenging because labeled data are scarce, the language is morphologically rich, and economic discourse shifts over time. We address these hurdles with a three-part framework. First,\n SSABE<\/jats:bold>\n a\n S<\/jats:italic>\n emi-\n S<\/jats:italic>\n upervised\n A<\/jats:italic>\n daptive\n B<\/jats:italic>\n oosting\n E<\/jats:italic>\n nsemble iteratively refines pseudo-labels, adjusts model weights by recent performance, and applies sector-aware voting to distill reliable labels from limited data. Second, the\n T<\/jats:italic>\n emporal\n S<\/jats:italic>\n entiment\n C<\/jats:italic>\n ontrastive\n M<\/jats:italic>\n odule (\n TSCM<\/jats:bold>\n ) aligns yearly embedding prototypes via contrastive loss, keeping the classifier robust against vocabulary drift and shifting economic regimes. Third,\n Temporal-SHAP<\/jats:bold>\n yields token-level attributions that reveal how term importance changes across years and industries, thereby making the system transparent to analysts. Evaluated on a 5-year (2018\u20132023) Bangla financial news corpus spanning eight sectors, our pipeline attains a macro-F\n 1<\/jats:sub>\n of 0.782 and 91.4 % explanation fidelity surpassing fine-tuned transformer and self-training baselines by 6 %\u201312 % absolute. Performance remains stable when labels are scarce, sectors are imbalanced, or economic shocks such as the inflation and currency decline of 2023 occur. Moreover, yearly sentiment scores and Temporal-SHAP attributions track inflation and exchange-rate trends, confirming real-world relevance. The proposed framework offers a scalable, interpretable solution for monitoring emerging-market news, supporting regulators, policymakers, and investors who rely on trustworthy Bangla-language insights.\n <\/jats:p>","DOI":"10.3389\/frai.2026.1724407","type":"journal-article","created":{"date-parts":[[2026,4,8]],"date-time":"2026-04-08T05:32:42Z","timestamp":1775626362000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":2,"title":["SSABE-TSCM: drift-aware and interpretable financial sentiment analysis for low-resource Bangla via adaptive semi-supervised and temporal contrastive modeling"],"prefix":"10.3389","volume":"9","author":[{"given":"Iftakhar Ali","family":"Khandokar","sequence":"first","affiliation":[{"name":"Department of Electrical and Computer Engineering, Marquette University","place":["Milwaukee, WI, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Priya","family":"Deshpande","sequence":"additional","affiliation":[{"name":"Department of Electrical and Computer Engineering, Marquette University","place":["Milwaukee, WI, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,4,8]]},"reference":[{"key":"B1","first-page":"43","article-title":"\u201cBanglanews: a Bangla news corpus for named entity recognition,\u201d","author":"Alam","year":"2020","journal-title":"Proceedings of the LREC 2020 Workshop on Resources for African Indigenous Languages (RAIL)"},{"key":"B2","article-title":"Finbert: financial sentiment analysis with pre-trained language models","author":"Araci","year":"2019","journal-title":"arXiv [preprint]"},{"key":"B3","article-title":"\u201cNeural machine translation by jointly learning to align and translate,\u201d","author":"Bahdanau","year":"2015","journal-title":"International Conference on Learning Representations (ICLR)"},{"key":"B4","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P19-1284","article-title":"\u201cInterpretable neural predictions with differentiable binary variables,\u201d","author":"Bastings","year":"2019","journal-title":"Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics"},{"key":"B5","article-title":"Banglabert: combating low-resource limitations with a pretrained language model for Bangla","author":"Bhattacharjee","year":"2021","journal-title":"arXiv [preprint]"},{"key":"B6","first-page":"29","article-title":"\u201cProcessing Bangla texts: the basics and beyond,\u201d","author":"Bhattacharya","year":"2005","journal-title":"Proceedings of the Second International Conference on Digital Information Management"},{"key":"B7","doi-asserted-by":"crossref","first-page":"92","DOI":"10.1145\/279943.279962","article-title":"\u201cCombining labeled and unlabeled data with co-training,\u201d","volume-title":"Proceedings of the Eleventh Annual Conference on Computational Learning Theory","author":"Blum","year":"1998"},{"key":"B8","first-page":"965","article-title":"\u201cTemporal adaptation of language models: an empirical study on social media text,\u201d","author":"Brandl","year":"2022","journal-title":"Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP)"},{"key":"B9","doi-asserted-by":"publisher","first-page":"74","DOI":"10.1109\/MIS.2017.4531228","article-title":"Sentiment analysis is a big suitcase","volume":"32","author":"Cambria","year":"2017","journal-title":"IEEE Intelligent Systems"},{"key":"B10","first-page":"1490","article-title":"\u201cKnowledge can be enriched explicitly: a method for temporal knowledge graph completion,\u201d","volume-title":"Proceedings of the 30th ACM International Conference on Information and Knowledge Management (CIKM)","author":"Cao","year":"2021"},{"key":"B11","first-page":"4310","article-title":"\u201cLexglue: a benchmark dataset for legal language understanding in English,\u201d","volume-title":"Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (ACL)","author":"Chalkidis","year":"2022"},{"key":"B12","first-page":"2145","article-title":"\u201cFintext: an interpretable and explainable financial text classification framework,\u201d","author":"Chen","year":"2022","journal-title":"Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing"},{"key":"B13","article-title":"\u201cSelective ensemble learning for neural text classification,\u201d","author":"Chen","year":"2020","journal-title":"COLING"},{"key":"B14","first-page":"276","article-title":"\u201cWhat does Bert look at? An analysis of Bert's attention,\u201d","volume-title":"BlackBoxNLP@ACL","author":"Clark","year":"2019"},{"key":"B15","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.aacl-main.46","article-title":"\u201cA survey of the state of explainable ai for natural language processing,\u201d","author":"Danilevsky","year":"2020","journal-title":"Proceedings of the 28th International Conference on Computational Linguistics"},{"key":"B16","first-page":"6514","article-title":"\u201cLeveraging uncertainty estimates for predicting classification accuracy,\u201d","author":"DeVries","year":"2018","journal-title":"Proceedings of the 32nd International Conference on Neural Information Processing Systems"},{"key":"B17","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.acl-main.408","article-title":"\u201cEraser: a benchmark for rationalizing NLP models,\u201d","author":"DeYoung","year":"2020","journal-title":"ACL 2020"},{"key":"B18","article-title":"\u201cTime-aware language models for clinical temporal question answering,\u201d","author":"Dhingra","year":"2021","journal-title":"EMNLP"},{"key":"B19","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1007\/3-540-45014-9_1","article-title":"\u201cEnsemble methods in machine learning,\u201d","volume-title":"International Workshop on Multiple Classifier Systems","author":"Dietterich","year":"2000"},{"key":"B20","first-page":"10785","article-title":"\u201cModeling temporal context for temporal relation extraction,\u201d","author":"Edmiston","year":"2022","journal-title":"Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP)"},{"key":"B21","first-page":"23","article-title":"\u201cA decision-theoretic generalization of on-line learning and an application to boosting,\u201d","volume-title":"European Conference on Computational Learning Theory","author":"Freund","year":"1997"},{"key":"B22","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.acl-main.51","article-title":"\u201cSimple, interpretable and stable method for detecting words with usage change across corpora,\u201d","author":"Gonen","year":"2020","journal-title":"Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (ACL)"},{"key":"B23","first-page":"445","article-title":"Resource constraints and progress in Bangla NLP: a survey","volume":"27","author":"Hasan","year":"2021","journal-title":"Nat. Lang. Eng"},{"key":"B24","first-page":"154","article-title":"\u201cBnhatespeech: a benchmark dataset for hate speech detection in Bengali,\u201d","author":"Hosain","year":"2021","journal-title":"Proceedings of the Fifth Workshop on Online Abuse and Harms"},{"key":"B25","first-page":"5281","article-title":"\u201cTime-aware language models as temporal knowledge bases,\u201d","author":"Huang","year":"2020","journal-title":"Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)"},{"key":"B26","first-page":"1","article-title":"\u201cTowards a sustainable NLP ecosystem for Bangla: Challenges and roadmap,\u201d","author":"Islam","year":"2021","journal-title":"ACM Transactions on Asian and Low-Resource Language Information Processing (TALLIP), Vol. 20"},{"key":"B27","doi-asserted-by":"publisher","first-page":"8018","DOI":"10.1609\/aaai.v34i05.6311","article-title":"Is Bert really robust? A strong baseline for natural language attack on text classification and entailment","volume":"34","author":"Jin","year":"2020","journal-title":"AAAI"},{"key":"B28","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.findings-emnlp.445","article-title":"\u201cIndicnlp suite: monolingual corpora, evaluation benchmarks and pre-trained multilingual language models for Indian languages,\u201d","author":"Kakwani","year":"2020","journal-title":"Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations"},{"key":"B29","first-page":"521","article-title":"\u201cSelf-supervised contrastive temporal modeling for language change detection,\u201d","author":"Ke","year":"2022","journal-title":"Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (ACL)"},{"key":"B30","first-page":"1384","article-title":"\u201cDiachronic word embeddings and semantic shifts: a survey,\u201d","author":"Kutuzov","year":"2018","journal-title":"Proceedings of the 27th International Conference on Computational Linguistics"},{"key":"B31","article-title":"\u201cTemporal ensembling for semi-supervised learning,\u201d","author":"Laine","year":"2017","journal-title":"International Conference on Learning Representations (ICLR)"},{"key":"B32","first-page":"6402","article-title":"Simple and scalable predictive uncertainty estimation using deep ensembles","volume":"30","author":"Lakshminarayanan","year":"2017","journal-title":"Adv. Neural Inf. Process. Syst"},{"key":"B33","article-title":"Mind the gap: assessing temporal generalization in neural language models","author":"Lazaridou","year":"2021","journal-title":"arXiv [preprint]"},{"key":"B34","article-title":"\u201cPseudo-label: The simple and efficient semi-supervised learning method for deep neural networks,\u201d","author":"Lee","year":"2013","journal-title":"Workshop on Challenges in Representation Learning, ICML"},{"key":"B35","first-page":"293","article-title":"News sentiment and its impact on stock volatility","volume":"52","author":"Li","year":"2014","journal-title":"J. Account. Res"},{"key":"B36","first-page":"4765","article-title":"\u201cA unified approach to interpreting model predictions,\u201d","author":"Lundberg","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"B37","doi-asserted-by":"publisher","first-page":"782","DOI":"10.1002\/asi.23062","article-title":"Good debt or bad debt: detecting semantic orientations in economic texts","volume":"65","author":"Malo","year":"2014","journal-title":"J. Am. Soc. Inf. Sci. Technol"},{"key":"B38","doi-asserted-by":"crossref","first-page":"70","DOI":"10.1145\/945645.945658","article-title":"\u201cSentiment analysis: capturing favorability using natural language processing,\u201d","volume-title":"Proceedings of the 2nd International Conference on Knowledge Capture","author":"Nasukawa","year":"2003"},{"key":"B39","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1561\/9781601981516","article-title":"Opinion mining and sentiment analysis","volume":"2","author":"Pang","year":"2008","journal-title":"Found. Trends Inf. Retr"},{"key":"B40","first-page":"1970","article-title":"\u201cInterpretable bow networks for sentence classification,\u201d","author":"Poerner","year":"2018","journal-title":"NAACL-HLT"},{"key":"B41","doi-asserted-by":"crossref","first-page":"1135","DOI":"10.1145\/2939672.2939778","article-title":"\u201cWhy should i trust you?\": explaining the predictions of any classifier,\u201d","volume-title":"Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining","author":"Ribeiro","year":"2016"},{"key":"B42","first-page":"678","article-title":"\u201cTime embeddings improve temporal understanding in language models,\u201d","author":"Rosin","year":"2022","journal-title":"Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies"},{"key":"B43","first-page":"3174","article-title":"\u201cBNLP: a toolkit for Bengali natural language processing,\u201d","author":"Sagor","year":"2020","journal-title":"Proceedings of the 12th Language Resources and Evaluation Conference"},{"key":"B44","doi-asserted-by":"publisher","first-page":"135","DOI":"10.1023\/A:1007649029923","article-title":"Boostexter: a boosting-based system for text categorization","volume":"39","author":"Schapire","year":"2000","journal-title":"Mach. Learn"},{"key":"B45","first-page":"596","article-title":"Fixmatch: simplifying semi-supervised learning with consistency and confidence","volume":"33","author":"Sohn","year":"2020","journal-title":"Adv. Neural Inf. Process. Syst"},{"key":"B46","first-page":"3319","article-title":"\u201cAxiomatic attribution for deep networks,\u201d","author":"Sundararajan","year":"2017","journal-title":"International Conference on Machine Learning"},{"key":"B47","first-page":"1195","article-title":"\u201cMean teachers are better role models: weight-averaged consistency targets improve semi-supervised deep learning results,\u201d","author":"Tarvainen","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"B48","doi-asserted-by":"publisher","first-page":"1139","DOI":"10.1111\/j.1540-6261.2007.01232.x","article-title":"Giving content to investor sentiment: The role of media in the stock market","volume":"62","author":"Tetlock","year":"2007","journal-title":"J. Financ"},{"key":"B49","article-title":"\u201cSemi-supervised text classification with heterogeneous graph co-training,\u201d","author":"Wang","year":"2022","journal-title":"ACL"},{"key":"B50","article-title":"\u201cELMO-TD: Fine-tuning ELMO with temporal drift modeling for social NLP,\u201d","author":"Wang","year":"2021","journal-title":"EMNLP Findings"},{"key":"B51","first-page":"6256","article-title":"Unsupervised data augmentation for consistency training","volume":"33","author":"Xie","year":"2020","journal-title":"Adv. Neural Inf. Process. Syst"},{"key":"B52","first-page":"2069","article-title":"\u201cFinqanet: fine-grained financial sentiment analysis using question answering-based language modeling,\u201d","author":"Yang","year":"2020","journal-title":"Proceedings of the 28th International Conference on Computational Linguistics"},{"key":"B53","article-title":"\u201cBert-of-theseus: compressing Bert by progressive module replacing,\u201d","author":"Zhou","year":"2020","journal-title":"EMNLP"},{"key":"B54","article-title":"\u201cTimebert: a pre-trained temporal language model for temporal information extraction,\u201d","author":"Zhou","year":"2022","journal-title":"ACL"},{"key":"B55","volume-title":"Semi-Supervised Learning Literature Survey","author":"Zhu","year":"2005"},{"key":"B56","first-page":"321","article-title":"\u201cLearning from labeled and unlabeled data with label propagation,\u201d","author":"Zhu","year":"2002","journal-title":"Advances in Neural Information Processing Systems"}],"container-title":["Frontiers in Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1724407\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,8]],"date-time":"2026-04-08T05:32:55Z","timestamp":1775626375000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1724407\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,8]]},"references-count":56,"alternative-id":["10.3389\/frai.2026.1724407"],"URL":"https:\/\/doi.org\/10.3389\/frai.2026.1724407","relation":{},"ISSN":["2624-8212"],"issn-type":[{"value":"2624-8212","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,4,8]]},"article-number":"1724407"}}