{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,10,10]],"date-time":"2025-10-10T01:47:01Z","timestamp":1760060821556,"version":"build-2065373602"},"reference-count":59,"publisher":"MDPI AG","issue":"10","license":[{"start":{"date-parts":[[2025,9,24]],"date-time":"2025-09-24T00:00:00Z","timestamp":1758672000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Shenzhen Stable Support Plan Program for Higher Education Institutions Research Program","award":["20231121164338004","24GSPCG14","72401200","72472103","72371168"],"award-info":[{"award-number":["20231121164338004","24GSPCG14","72401200","72472103","72371168"]}]},{"name":"High-Level Achievements Cultivation Project of the Third Phase of High-Level University Construction of Shenzhen University","award":["20231121164338004","24GSPCG14","72401200","72472103","72371168"],"award-info":[{"award-number":["20231121164338004","24GSPCG14","72401200","72472103","72371168"]}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["20231121164338004","24GSPCG14","72401200","72472103","72371168"],"award-info":[{"award-number":["20231121164338004","24GSPCG14","72401200","72472103","72371168"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Systems"],"abstract":"<jats:p>The textual analysis of Management Discussion and Analysis (MD&amp;A) reveals valuable insights into corporate operational performance and future risks. However, techniques for accurately extracting sentiment from unstructured Chinese MD&amp;A texts still lack comprehensiveness. Existing studies related to sentiment analysis often use lexicon-based methods, which rely on predefined, context-agnostic word lists and accurate Chinese word segmentation and struggle with domain-specific terminology, leading to limited accuracy and interpretability. Although research has attempted to develop context-aware lexicons and language models, these methods still face limitations when applied to long and complex financial texts. To address the limitations, we propose MDARisk, a novel framework for corporate misconduct prediction. The core of MDARisk is a MultiSenti module, which leverages a multi-agent LLM approach to extract comprehensive and contextual sentiment from MD&amp;A. Unlike lexicon methods, our LLM-based module interprets words based on their surrounding semantic context, allowing it to decipher nuanced expressions and specialized financial language. We first conduct an econometric validation using fixed-effects logit models to test whether the MultiSenti-derived MD&amp;A sentiment is significantly associated with subsequent corporate misconduct. We then evaluate out-of-sample predictive utility by adding this sentiment feature to multiple classifiers and assessing its incremental gains over the baseline model. Empirical results demonstrate that our approach provides a more reliable sentiment-based indicator for misconduct risk, achieves higher predictive accuracy, and outperforms the traditional financial sentiment analysis approach. Our MDARisk framework provides a cost-efficient approach for automated disclosure screening, benefiting auditors, regulators, and investors in assessing potential misconduct risks.<\/jats:p>","DOI":"10.3390\/systems13100839","type":"journal-article","created":{"date-parts":[[2025,9,24]],"date-time":"2025-09-24T13:16:11Z","timestamp":1758719771000},"page":"839","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["LLM-Driven Sentiment Analysis in MD&amp;A: A Multi-Agent Framework for Corporate Misconduct Prediction"],"prefix":"10.3390","volume":"13","author":[{"given":"Yeling","family":"Liu","sequence":"first","affiliation":[{"name":"College of Economics, Shenzhen University, Shenzhen 518055, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0001-7380-681X","authenticated-orcid":false,"given":"Yongkang","family":"Liu","sequence":"additional","affiliation":[{"name":"College of Economics, Shenzhen University, Shenzhen 518055, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Kai","family":"Yang","sequence":"additional","affiliation":[{"name":"College of Economics, Shenzhen University, Shenzhen 518055, China"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1968","published-online":{"date-parts":[[2025,9,24]]},"reference":[{"key":"ref_1","first-page":"699","article-title":"Pressure, Opportunity and Predisposition: A Multivariate Model of Corporate Illegality","volume":"20","author":"Baucus","year":"1994","journal-title":"J. Manag."},{"key":"ref_2","doi-asserted-by":"crossref","unstructured":"Ichev, R. (2023). Reported Corporate Misconducts: The Impact on the Financial Markets. PLoS ONE, 18.","DOI":"10.1371\/journal.pone.0276637"},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"539","DOI":"10.1007\/s10551-016-3234-3","article-title":"Social Trust and Corporate Misconduct: Evidence from China","volume":"151","author":"Dong","year":"2018","journal-title":"J. Bus. Ethics"},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"221","DOI":"10.1016\/j.jacceco.2008.02.003","article-title":"Annual Report Readability, Current Earnings, and Earnings Persistence","volume":"45","author":"Li","year":"2008","journal-title":"J. Account. Econ."},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"101299","DOI":"10.1016\/j.jacceco.2020.101299","article-title":"The Spillover Effects of MD&A Disclosures for Real Investment: The Role of Industry Competition","volume":"70","author":"Durnev","year":"2020","journal-title":"J. Account. Econ."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"1643","DOI":"10.1111\/jofi.12162","article-title":"Measuring Readability in Financial Disclosures","volume":"69","author":"Loughran","year":"2014","journal-title":"J. Financ."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"605","DOI":"10.22300\/0896-5803.41.4.605","article-title":"On the Relationship between Market Sentiment and Commercial Real Estate Performance?A Textual Analysis Examination","volume":"41","author":"Beracha","year":"2019","journal-title":"J. Real Estate Res."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"137","DOI":"10.1007\/s10551-022-05120-2","article-title":"Using Machine Learning to Predict Corporate Fraud: Evidence Based on the GONE Framework","volume":"186","author":"Xu","year":"2023","journal-title":"J. Bus. Ethics"},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"122266","DOI":"10.1016\/j.jclepro.2020.122266","article-title":"Detecting Corporate Misconduct through Random Forest in China\u2019s Construction Industry","volume":"268","author":"Wang","year":"2020","journal-title":"J. Clean. Prod."},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"7034","DOI":"10.1287\/mnsc.2021.4211","article-title":"Tone at the Bottom: Measuring Corporate Misconduct Risk from the Text of Employee Reviews","volume":"68","author":"Campbell","year":"2022","journal-title":"Manag. Sci."},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"Bel, N., Bracons, G., and Anderberg, S. (2021). Finding Evidence of Fraudster Companies in the CEO\u2019s Letter to Shareholders with Sentiment Analysis. Information, 12.","DOI":"10.3390\/info12080307"},{"key":"ref_12","first-page":"363","article-title":"Are Investors Influenced By How Earnings Press Releases Are Written?","volume":"45","author":"Henry","year":"2008","journal-title":"J. Bus. Commun. 1973"},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"35","DOI":"10.1111\/j.1540-6261.2010.01625.x","article-title":"When Is a Liability Not a Liability? Textual Analysis, Dictionaries, and 10-Ks","volume":"66","author":"Loughran","year":"2011","journal-title":"J. Financ."},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"17","DOI":"10.1016\/j.artmed.2015.03.006","article-title":"Sentiment Analysis in Medical Settings: New Opportunities and Challenges","volume":"64","author":"Denecke","year":"2015","journal-title":"Artif. Intell. Med."},{"key":"ref_15","first-page":"23","article-title":"Forecasting financial distress of listed companies with textual content of the information disclosure: A study based MD&A in Chinese annual reports","volume":"27","author":"Chen","year":"2019","journal-title":"Chin. J. Manag. Sci."},{"key":"ref_16","first-page":"7223","article-title":"ChatGPT for Textual Analysis? How to Use Generative LLMs in Accounting Research","volume":"71","year":"2025","journal-title":"Manag. Sci."},{"key":"ref_17","first-page":"1323","article-title":"Media textual sentiment and Chinese stock return predictability","volume":"21","author":"Jiang","year":"2021","journal-title":"China Econ. Q."},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"353","DOI":"10.1007\/s11301-021-00244-7","article-title":"The Link between Corporate Governance and Corporate Financial Misconduct. A Review of Archival Studies and Implications for Future Research","volume":"73","author":"Velte","year":"2023","journal-title":"Manag. Rev. Q."},{"key":"ref_19","first-page":"2538","article-title":"Board Independence and Corporate Misconduct: A Cross-National Meta-Analysis","volume":"45","author":"Neville","year":"2019","journal-title":"J. Manag."},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"732","DOI":"10.1007\/s11142-017-9435-x","article-title":"Financial Reporting Fraud and Other Forms of Misconduct: A Multidisciplinary Review of the Literature","volume":"23","author":"Amiram","year":"2018","journal-title":"Rev. Acc. Stud."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Lou, Y.-I., and Wang, M.-L. (2011). Fraud Risk Factor Of The Fraud Triangle Assessing The Likelihood Of Fraudulent Financial Reporting. JBER, 7.","DOI":"10.19030\/jber.v7i2.2262"},{"key":"ref_22","unstructured":"Beasley, M., Carcello, J., Hermanson, D., and Committee of Sponsoring Organizations of the Treadway Commission (2025, September 11). Fraudulent Financial Reporting: 1987\u20131997: An Analysis of U.S. Public Companies: Research Report. Association Sections, Divisions, Boards, Teams 1999. Available online: https:\/\/egrove.olemiss.edu\/aicpa_assoc\/249\/."},{"key":"ref_23","first-page":"1","article-title":"Multiple Directorships and Corporate Misconduct: The Moderating Influences of Board Size and Outside Directors","volume":"25","author":"Schnake","year":"2008","journal-title":"J. Bus. Strateg."},{"key":"ref_24","doi-asserted-by":"crossref","unstructured":"(2011). Aiyesha Dey; Ellen Engel; Xiaohui Liu CEO and Board Chair Roles: To Split or Not to Split?. J. Corp. Financ., 17, 1595\u20131618.","DOI":"10.1016\/j.jcorpfin.2011.09.001"},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"45","DOI":"10.5465\/amj.2010.48036912","article-title":"CEO Hubris and Firm Risk Taking in China: The Moderating Role of Managerial Discretion","volume":"53","author":"Li","year":"2010","journal-title":"AMJ"},{"key":"ref_26","first-page":"89","article-title":"Can a short selling mechanism restrain corporate fraud?","volume":"54","author":"Meng","year":"2019","journal-title":"Econ. Res."},{"key":"ref_27","doi-asserted-by":"crossref","unstructured":"Zou, Y., Zhang, R., Meng, Q., and Hou, D. (2019). Can stock market liberalization restrain corporate fraud? Evidence from \u201cShang-hai-Hong Kong stock connect\u201d scheme. China Soft Sci., 120\u2013134.","DOI":"10.2139\/ssrn.3483638"},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"164","DOI":"10.1016\/j.dss.2010.07.012","article-title":"Making Words Work: Using Financial Text as a Predictor of Financial Events","volume":"50","author":"Cecchini","year":"2010","journal-title":"Decis. Support Syst."},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"197","DOI":"10.1080\/2573234X.2024.2342773","article-title":"Predicting Fraud in MD&A Sections Using Deep Learning","volume":"7","author":"Skillicorn","year":"2024","journal-title":"J. Bus. Anal."},{"key":"ref_30","doi-asserted-by":"crossref","first-page":"1267","DOI":"10.1111\/jofi.12027","article-title":"Sentiment during Recessions","volume":"68","year":"2013","journal-title":"J. Financ."},{"key":"ref_31","first-page":"69","article-title":"Listed Companies\u2019 Annual Report Incremental Text Information and Fraud Risk Prediction: From the Perspective of Tone","volume":"39","author":"Guo","year":"2022","journal-title":"Stat. Res."},{"key":"ref_32","doi-asserted-by":"crossref","first-page":"1187","DOI":"10.1111\/1475-679X.12123","article-title":"Textual Analysis in Accounting and Finance: A Survey","volume":"54","author":"Loughran","year":"2016","journal-title":"J. Account. Res."},{"key":"ref_33","doi-asserted-by":"crossref","first-page":"129","DOI":"10.1111\/fire.12280","article-title":"Do Changes in MD&A Section Tone Predict Investment Behavior?","volume":"57","author":"Berns","year":"2022","journal-title":"Financ. Rev."},{"key":"ref_34","doi-asserted-by":"crossref","first-page":"992","DOI":"10.1016\/j.jbankfin.2011.10.013","article-title":"Bliss Earnings Conference Calls and Stock Returns: The Incremental Informativeness of Textual Tone","volume":"36","author":"Price","year":"2012","journal-title":"J. Bank. Financ."},{"key":"ref_35","doi-asserted-by":"crossref","first-page":"1049","DOI":"10.1111\/j.1475-679X.2010.00382.x","article-title":"The Information Content of Forward-Looking Statements in Corporate Filings\u2014A Na\u00efve Bayesian Machine Learning Approach","volume":"48","author":"Li","year":"2010","journal-title":"J. Account. Res."},{"key":"ref_36","doi-asserted-by":"crossref","first-page":"139","DOI":"10.1016\/j.knosys.2017.05.001","article-title":"Mining Corporate Annual Reports for Intelligent Detection of Financial Statement Fraud\u2014A Comparative Study of Machine Learning Methods","volume":"128","author":"Hajek","year":"2017","journal-title":"Knowl. Based Syst."},{"key":"ref_37","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1108\/CFRI-08-2019-0134","article-title":"Textual Analysis for China\u2019s Financial Markets: A Review and Discussion","volume":"10","author":"Huang","year":"2019","journal-title":"China Financ. Rev. Int."},{"key":"ref_38","doi-asserted-by":"crossref","first-page":"199","DOI":"10.1111\/1475-679X.12292","article-title":"Detecting Accounting Fraud in Publicly Traded, U.S. Firms Using a Machine Learning Approach","volume":"58","author":"Bao","year":"2020","journal-title":"J. Account. Res."},{"key":"ref_39","doi-asserted-by":"crossref","unstructured":"Bello, A., Ng, S.-C., and Leung, M.-F. (2023). A BERT Framework to Sentiment Analysis of Tweets. Sensors, 23.","DOI":"10.3390\/s23010506"},{"key":"ref_40","doi-asserted-by":"crossref","first-page":"105837","DOI":"10.1016\/j.frl.2024.105837","article-title":"Unveiling Tone Manipulation in MD&A: Evidence from ChatGPT Experiments","volume":"67","author":"Song","year":"2024","journal-title":"Financ. Res. Lett."},{"key":"ref_41","first-page":"20","article-title":"Do management tones help to forecast firms\u2019 future performance: A textual analysis based on annual earnings communication conferences of listed companies in China","volume":"2","author":"Xie","year":"2015","journal-title":"Account. Res."},{"key":"ref_42","doi-asserted-by":"crossref","first-page":"1083","DOI":"10.2308\/accr-50684","article-title":"Tone Management","volume":"89","author":"Huang","year":"2014","journal-title":"Account. Rev."},{"key":"ref_43","first-page":"99","article-title":"Corporate Fraud and Manipulation of Annual Report Text Information","volume":"3","author":"Tan","year":"2022","journal-title":"China Soft Sci."},{"key":"ref_44","doi-asserted-by":"crossref","first-page":"12801","DOI":"10.1007\/s11227-021-03709-4","article-title":"Intelligent Sentinet-Based Lexicon for Context-Aware Sentiment Analysis: Optimized Neural Network for Sentiment Classification on Social Media","volume":"77","author":"Uma","year":"2021","journal-title":"J. Supercomput."},{"key":"ref_45","unstructured":"Moradi-Kamali, H., Rajabi-Ghozlou, M.-H., Ghazavi, M., Soltani, A., Sattarzadeh, A., and Entezari-Maleki, R. (2025). Market-Derived Financial Sentiment Analysis: Context-Aware Language Models for Crypto Forecasting. arXiv."},{"key":"ref_46","unstructured":"Araci, D. (May, January 26). FinBERT: Financial sentiment analysis with pre-trained language models. Proceedings of the Eleventh International Conference on Learning Representations (ICLR 2020), Addis Ababa, Ethiopia."},{"key":"ref_47","doi-asserted-by":"crossref","first-page":"166","DOI":"10.1007\/s10462-025-11162-5","article-title":"BERT Applications in Natural Language Processing: A Review","volume":"58","author":"Gardazi","year":"2025","journal-title":"Artif. Intell. Rev."},{"key":"ref_48","doi-asserted-by":"crossref","first-page":"78","DOI":"10.1111\/1911-3846.12150","article-title":"Corporate Political Connections and Tax Aggressiveness","volume":"33","author":"Kim","year":"2016","journal-title":"Contemp. Account. Res."},{"key":"ref_49","doi-asserted-by":"crossref","first-page":"75","DOI":"10.2307\/25148625","article-title":"Design Science in Information Systems Research","volume":"28","author":"Hevner","year":"2004","journal-title":"MIS Q."},{"key":"ref_50","unstructured":"Wang, X., Wei, J., Schuurmans, D., Le, Q., Chi, E., Narang, S., Chowdhery, A., and Zhou, D. (2023, January 1\u20135). Self-consistency improves chain of thought reasoning in language models. Proceedings of the Eleventh International Conference on Learning Representations (ICLR 2023), Kigali, Rwanda."},{"key":"ref_51","doi-asserted-by":"crossref","first-page":"17","DOI":"10.1111\/j.1911-3846.2010.01041.x","article-title":"Predicting Material Accounting Misstatements","volume":"28","author":"Dechow","year":"2011","journal-title":"Contemp. Account. Res."},{"key":"ref_52","first-page":"443","article-title":"An Empirical Analysis of the Relation between the Board of Director Composition and Financial Statement Fraud","volume":"71","author":"Beasley","year":"1996","journal-title":"Account. Rev."},{"key":"ref_53","unstructured":"Guyon, I., and Elisseeff, A. (2003). An introduction to variable and feature selection. J. Mach. Learn. Res., 1157\u20131182."},{"key":"ref_54","doi-asserted-by":"crossref","first-page":"321","DOI":"10.1613\/jair.953","article-title":"SMOTE: Synthetic Minority Over-Sampling Technique","volume":"16","author":"Chawla","year":"2002","journal-title":"J. Artif. Intell. Res."},{"key":"ref_55","doi-asserted-by":"crossref","unstructured":"Saito, T., and Rehmsmeier, M. (2015). The Precision-Recall Plot Is More Informative than the ROC Plot When Evaluating Binary Classifiers on Imbalanced Datasets. PLoS ONE, 10.","DOI":"10.1371\/journal.pone.0118432"},{"key":"ref_56","doi-asserted-by":"crossref","unstructured":"Davis, J., and Goadrich, M. (2006). The Relationship between Precision-Recall and ROC Curves. Proceedings of the 23rd international conference on Machine learning, Pittsburgh, PA, USA, 25\u201329 June 2025, Association for Computing Machinery.","DOI":"10.1145\/1143844.1143874"},{"key":"ref_57","doi-asserted-by":"crossref","first-page":"29","DOI":"10.1016\/S0929-1199(01)00039-6","article-title":"Institutional Monitoring and Opportunistic Earnings Management","volume":"8","author":"Chung","year":"2002","journal-title":"J. Corp. Financ."},{"key":"ref_58","doi-asserted-by":"crossref","first-page":"137","DOI":"10.1093\/oxfordjournals.pan.a004868","article-title":"Logistic Regression in Rare Events Data","volume":"9","author":"King","year":"2001","journal-title":"Political Anal."},{"key":"ref_59","first-page":"4129","article-title":"A Comparative Analysis of Machine Learning Techniques for Credit Scoring","volume":"15","author":"Nwulu","year":"2012","journal-title":"Information"}],"container-title":["Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2079-8954\/13\/10\/839\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,9]],"date-time":"2025-10-09T18:49:03Z","timestamp":1760035743000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2079-8954\/13\/10\/839"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,9,24]]},"references-count":59,"journal-issue":{"issue":"10","published-online":{"date-parts":[[2025,10]]}},"alternative-id":["systems13100839"],"URL":"https:\/\/doi.org\/10.3390\/systems13100839","relation":{},"ISSN":["2079-8954"],"issn-type":[{"type":"electronic","value":"2079-8954"}],"subject":[],"published":{"date-parts":[[2025,9,24]]}}}