Large Language Models for analyzing project stakeholders’ environmental opinions to support decision-making

Abstract

In the global context of sustainable development, stakeholder concerns about the environmental impacts of infrastructure projects have become increasingly prominent, which can significantly influence the progress of projects. However, integrating changing environmental opinions into project decision-making remains a challenge due to the complexity, highly dynamic nature and volume of data. Large Language Models (LLMs) have emerged as transformative tools for efficiently and rapidly analyzing this type of data, offering new opportunities for enhancing decision-making processes. This research proposes a framework utilizing LLM for three major approaches in opinion analysis among stakeholders: sentiment analysis, stance analysis, and topic modeling. The framework has been applied to the case of the Scarborough Gas Project in Western Australia. A set of smaller models, including Neural Networks (NNs), Support Vector Machines (SVMs), Random Forest, Logistic Regression, and BERT, were fine-tuned using GPT-3.5 as a base and compared for performance in sentiment and stance analysis, with SVM achieving the highest accuracy rates of 83.90% and 87.55%, respectively. Integrating LLMs into topic modeling also significantly enhanced the interpretation of stakeholder environmental opinions by transforming keyword lists generated by traditional LDA methods into coherent narratives, reducing reliance on human interpretation, refining themes, and enabling a more comprehensive understanding of environmental, political, and legal issues. This study presents the first unified framework that integrates LLM embeddings with external classifiers to simultaneously analyze all three analytical tasks, to our knowledge. Central to the framework is the theoretically grounded Sentiment–Stance–Topic Matrix and Decision-Making Map, which systematically translate unstructured stakeholder input into prioritized engagement actions. By categorizing sentiment, stance, and topic configurations into targeted strategies, the framework offers structured, data-driven guidance for project decision-makers. This approach bridges gaps in traditional stakeholder analysis and provides a transferable decision-support tool, enabling more inclusive, responsive project governance aligned with global sustainable development goals.

Graphical abstract

Keywords

online stakeholder engagement / large language models / decision-making / sentiment analysis / topic modeling / environmental discourse

References

[1] Abdelrazek AEid YGawish EMedhat WHassan A (2023). Topic modeling algorithms and applications: A survey. Information Systems, 112, 102176

[2] Alam S, Yao N, (2019). The impact of preprocessing steps on the accuracy of machine learning algorithms in sentiment analysis. Computational & Mathematical Organization Theory, 25( 3): 319–335

[3] ALDayel A, Magdy W, (2021). Stance detection on social media: State of the art and trends. Information Processing & Management, 58( 4): 102597

[4] Aljebreen M, Alabduallah B, Asiri M M, Salama A S, Assiri M, Ibrahim S S, (2023). Moth flame optimization with hybrid deep learning based sentiment classification toward ChatGPT on Twitter. IEEE Access: Practical Innovations, Open Solutions, 11: 104984–104991

[5] Amangeldi D, Usmanova A, Shamoi P, (2024). Understanding environmental posts: Sentiment and emotion analysis of social media data. IEEE Access, 12: 33504–33523

[6] Andrus B R, Nasiri Y, Cui S, Cullen B, Fulda N, (2022). Enhanced story comprehension for large language models through dynamic document-based knowledge graphs. In: Proceedings of the AAAI Conference on Artificial Intelligence, 36( 10): 10436–10444

[7] Bello A, Ng S C, Leung M F, (2023). A BERT framework to sentiment analysis of tweets. Sensors, 23( 1): 1–1

[8] Blei D MNg A YJordan M ILafferty J (2003). Latent dirichlet allocation. Journal of Machine Learning Research EBSCOhost. Available at the website of ebsco.com

[9] Breiman L, (2001). Random Forests. Machine Learning, 45( 1): 5–32

[10] Brown T B, Mann B, Ryder N, Subbiah M, Kaplan J, Dhariwal P, Neelakantan A, Shyam P, Sastry G, Askell A, Agarwal S, Herbert-Voss A, Krueger G, Henighan T, Child R, Ramesh A, Ziegler D M, Wu J, Winter C, Hesse C, Chen M, Sigler E, Litwin M, Gray S, Chess B, Clark J, Berner C, McCandlish S, Radford A, Sutskever I, Amodei D, (2020). Language models are few-shot learners. Advances in Neural Information Processing Systems, 33: 1877–1901

[11] Bu Z, Liu J, Liu J, (2025). Impact of emotions on the behavioral strategies of PPP project stakeholders: An RDEU evolutionary game analysis. Engineering, Construction, and Architectural Management, 32( 4): 2242–2271

[12] Bucur C, Tudorica B, Andrei J V, Dusmanescu D, Paraschiv D, Teodor C, (2024). Sentiment analysis of global news on environmental issues: Insights into public perception and its impact on low-carbon economy transition. Frontiers in Environmental Science, 12: 1360304

[13] Bucy E P, Evans H K, (2022). Media-centric and politics-centric views of media and democracy: A longitudinal analysis of political communication and the international journal of press/politics. Political Communication, 39( 2): 254–265

[14] Burke J J, Clark C E, (2016). The business case for integrated reporting: Insights from leading practitioners, regulators, and academics. Business Horizons, 59( 3): 273–283

[15] Buysse K, Verbeke A, (2003). Proactive environmental strategies: A stakeholder management perspective. Strategic Management Journal, 24( 5): 453–470

[16] Büyüközkan G, Karabulut Y, (2017). Energy project performance evaluation with sustainability perspective. Energy, 119: 549–560

[17] Chung K S K, Eskerod P, Jepsen A L, Zhang J, (2023). Response strategies for community stakeholder engagement on social media: A case study of a large infrastructure project. International Journal of Project Management, 41( 5): 102495. Q1

[18] Corbett J, Savarimuthu B T R, (2022). From tweets to insights: A social media analysis of the emotion discourse of sustainable energy in the united states. Energy Research & Social Science, 89: 102515

[19] Cortes C, Vapnik V, (1995). Support-vector networks. Machine Learning, 20( 3): 273–297

[20] Cundy A B, Bardos R P, Church A, Puschenreiter M, Friesl-Hanl W, Müller I, Neu S, Mench M, Witters N, Vangronsveld J, (2013). Developing principles of sustainability and stakeholder engagement for “gentle” remediation approaches: The European context. Journal of Environmental Management, 129: 283–291

[21] Cuppen E, (2018). The value of social conflicts. Critiquing invited participation in energy projects. Energy Research & Social Science, 38: 28–32

[22] Daemi A, Chugh R, Kanagarajoo M V, (2021). Social media in project management: A systematic narrative literature review. International Journal of Information Systems and Project Management, 8( 4): 5–20

[23] Das R, Singh T D, (2023). Multimodal sentiment analysis: A survey of methods, trends, and challenges. ACM Computing Surveys, 55( 13s): 1–38

[24] Devine-Wright P, (2011). Public engagement with large-scale renewable energy technologies: Breaking the cycle of NIMBYism. Wiley Interdisciplinary Reviews: Climate Change, 2( 1): 19–26

[25] Devlin JChang M WLee KToutanova K (2019). BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In J. Burstein, C. Doran, & T. Solorio (Eds), In: Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1. Association for Computational Linguistics, 4171–4186

[26] Dogra V., Verma S., Kavita M., Woźniak J., Shafi M F, (2024). Shortcut learning explanations for deep natural language processing: A survey on dataset Biases. IEEE Access, 12: 26183–26195

[27] Dokshin F A, (2021). Variation of public discourse about the impacts of fracking with geographic scale and proximity to proposed development. Nature Energy, 6( 10): 961

[28] Druckman J N, Levendusky M S, (2019). What do we measure when we measure affective polarization. Public Opinion Quarterly, 83( 1): 114–122

[29] Effrosynidis D, Sylaios G, Arampatzis A, (2022). Exploring climate change on Twitter using seven aspects: Stance, sentiment, aggressiveness, temperature, gender, topics, and disasters. PLoS One, 17( 9): e0274213

[30] Elmitwalli S, Mehegan J, (2024). Sentiment analysis of COP9-related tweets: A comparative study of pre-trained models and traditional techniques. Frontiers in Big Data, 7: 1357926

[31] Elsawah S, Hamilton S, Jakeman A, Rothman D, Schweizer V, Trutnevyte E, Carlsen H, Drakes C, Frame B, Fu B, Guivarch C, Haasnoot M, Kemp-Benedict E, Kok K, Kosow H, Ryan M, van Delden H, (2020). Scenario processes for socio-environmental systems analysis of futures: A review of recent efforts and a salient research agenda for supporting decision making. Science of the Total Environment, 729: 138393

[32] Engert S, Rauter R, Baumgartner R J, (2016). Exploring the integration of corporate sustainability into strategic management: A literature review. Journal of Cleaner Production, 112: 2833–2850

[33] Ershadi M, Goodarzi F, (2021). Core capabilities for achieving sustainable construction project management. Sustainable Production and Consumption, 28: 1396–1410

[34] Freeman R E, Dmytriyev S D, Phillips R A, (2021). Stakeholder theory and the resource-based view of the firm. Journal of Management, 47( 7): 1757–1770

[35] Frooman J, (1999). Stakeholder influence strategies. Academy of Management Review, 24( 2): 191–205

[36] Fu L, Tu X, Liao J, (2024). Asymmetric impacts of climate policy uncertainty, investor sentiment on energy prices and renewable energy consumption: Evidence from NARDL and wavelet coherence. Journal of Environmental Management, 367: 122057

[37] Fulton E, Boschetti F, Sporcic M, Jones T, Little L R, Dambacher J, Gray R, Scott R, Gorton R, (2015). A multi-model approach to engaging stakeholder and modellers in complex environmental problems. Environmental Science & Policy, 48: 44–56

[38] Gao X, Zeng S, Zeng R, Shi J J, Song R, (2022). Multiple-stakeholders’ game and decision-making behaviors in green management of megaprojects. Computers & Industrial Engineering, 171: 108392

[39] Goutte CGaussier E (2005). A Probabilistic Interpretation of Precision, Recall and F-Score, with Implication for Evaluation. In D. E. Losada & J. M. Fernández-Luna (Eds). Advances in Information Retrieval, Springer. 345–359

[40] Harrin E (2021). Social Media for Project Managers. Project Management Institute

[41] He L, Yin T, Zheng K, (2022). They may not work! An evaluation of eleven sentiment analysis tools on seven social media datasets. Journal of Biomedical Informatics, 132: 104142–104153

[42] Holifield R, Williams K C, (2019). Recruiting, integrating, and sustaining stakeholder participation in environmental management: A case study from the great lakes areas of concern. Journal of Environmental Management, 230: 422–433

[43] Huang J, Ling C X, (2005). Using AUC and accuracy in evaluating learning algorithms. IEEE Transactions on Knowledge and Data Engineering, 17( 3): 299–310

[44] Hysa B, Spalek S, (2019). Opportunities and threats presented by social media in project management. Heliyon, 5( 4): e01488

[45] Jaiswal A, Samuel C, Kadabgaon V M, (2018). Statistical trend analysis and forecast modeling of air pollutants. Global Journal of Environmental Science and Management, 4: 427–438

[46] Jelodar H, Wang Y, Yuan C, Feng X, Jiang X, Li Y, Zhao L, (2019). Latent Dirichlet allocation (LDA) and topic modeling: Models, applications, a survey. Multimedia Tools and Applications, 78( 11): 15169–15211

[47] Jeong D, Hwang S, Kim J, Yu H, Park E, (2023). Public perspective on renewable and other energy resources: Evidence from social media big data and sentiment analysis. Energy Strategy Reviews, 50: 101243

[48] Jiang H, Qiang M, Lin P, (2016). Finding academic concerns of the Three Gorges Project based on a topic modeling approach. Ecological Indicators, 60: 693–701

[49] Jung H S, Lee H, Woo Y S, Baek S Y, Kim J H, (2024). Expansive data, extensive model: Investigating discussion topics around LLM through unsupervised machine learning in academic papers and news. PLoS One, 19( 5): e0304680

[50] Khanra S, Dhir A, Kaur P, Mäntymäki M, (2021). Bibliometric analysis and literature review of ecotourism: Toward sustainable development. Tourism Management Perspectives, 37: 100777

[51] Ki C W, Chong S M, Ha-Brookshire J E, (2020). How fashion can achieve sustainable development through a circular economy and stakeholder engagement: A systematic literature review. Corporate Social Responsibility and Environmental Management, 27( 6): 2401–2424

[52] Kim S, Kim K, Wonjeong Jo C W, (2024). Accuracy of a large language model in distinguishing anti- and pro-vaccination messages on social media: The case of human papillomavirus vaccination. Preventive Medicine Reports, 42: 102723

[53] Küçük D, Can F, (2020). Stance detection: A survey. ACM Computing Surveys, 53( 1): 1–37

[54] Kwon O H, Vu K, Bhargava N, Radaideh M I, Cooper J, Joynt V, Radaideh M I, (2024). Sentiment analysis of the United States public support of nuclear power on social media using large language models. Renewable & Sustainable Energy Reviews, 200: 114570

[55] Lai Y, Kontokosta C E, (2019). Topic modeling to discover the thematic structure and spatial-temporal patterns of building renovation and adaptive reuse in cities. Computers, Environment and Urban Systems, 78: 101383

[56] Lam J, Cheung L, Wang S, Li V, (2019). Stakeholder concerns of air pollution in hong kong and policy implications: A big-data computational text analysis approach. Environmental Science & Policy, 101: 374–382

[57] Laureate C D P, Buntine W, Linger H, (2023). A systematic review of the use of topic models for short text social media analysis. Artificial Intelligence Review, 56( 12): 14223–14255

[58] Lewandowska-Tomaszczyk BLiebeskind C (2024). Opinion events and stance types: Advances in LLM performance with ChatGPT and Gemini. Lodz Papers in Pragmatics

[59] Li Y, Frans V F, Song Y, Cai M, Zhang Y, Liu J, (2023). SDGdetector: An R-based text mining tool for quantifying efforts toward Sustainable Development Goals. Journal of Open Source Software, 8( 84): 5124

[60] Lu W, Luu R K, Buehler M J, (2025). Fine-tuning large language models for domain adaptation: Exploration of training strategies, scaling, model merging and synergistic capabilities. Npj Computational Materials, 11( 1): 84

[61] Ma Z, Mir R, Dalton C A, Godfrey K E, (2024). Choosing appropriate regularization parameters by splitting data into training and validation sets—Application in global surface-wave tomography. Seismological Research Letters, 95( 5): 3029–3041

[62] Mäntylä M V, Graziotin D, Kuutila M, (2018). The evolution of sentiment analysis—A review of research topics, venues, and top cited papers. Computer Science Review, 27: 16–32

[63] Marquart-Pyatt S, (2015). Public opinion about the environment: Testing measurement equivalence across countries. International Journal of Sociology, 45( 4): 309–326

[64] Mathew E S, Tembely M, AlAmeri W, Al-Shalabi E W, Shaik A R, (2021). Artificial intelligence coreflooding simulator for special core data analysis. SPE Reservoir Evaluation & Engineering, 24( 04): 780–808

[65] Miah M S U, Kabir M M, Sarwar T B, Safran M, Alfarhood S, Mridha M F, (2024). A multimodal approach to cross-lingual sentiment analysis with ensemble of transformer and LLM. Scientific Reports, 14( 1): 9603

[66] Morsing M, Spence L J, (2019). Corporate social responsibility (CSR) communication and small and medium sized enterprises: The governmentality dilemma of explicit and implicit CSR communication. Human Relations, 72( 12): 1920–1947

[67] Navigli R, Conia S, Ross B, (2023). Biases in Large Language Models: Origins, inventory, and discussion. Journal of Data and Information Quality, 15( 2): 10:1–10:21

[68] Olander S, Landin A, (2008). A comparative study of factors affecting the external stakeholder management process. Construction Management and Economics, 26( 6): 553–561

[69] Oyewole M O, Komolafe M O, Gbadegesin J T, (2023). Understanding stakeholders’ opinion and willingness on the adoption of sustainable residential property features in a developing property market. International Journal of Construction Management, 23( 2): 358–370

[70] Pang B, Lee L, (2008). Opinion mining and sentiment analysis. Foundations and Trends® in Information Retrieval, 2( 1–2): 1–135

[71] Perlaviciute G, Steg L, Contzen N, Roeser S, Huijts N, (2018). Emotional responses to energy projects: Insights for responsible decision making in a sustainable energy transition. Sustainability, 10( 7): 2526

[72] Praveena S, Aris A, (2021). The impacts of COVID-19 on the environmental sustainability: A perspective from the southeast asian region. Environmental Science and Pollution Research International, 28( 45): 63829–63836

[73] Project Management Institute (2022). Guide to the Project Management Body of Knowledge (PMBOK guide) and the Standard for project management (7th ed edition). Project Management Institute

[74] Raineri N, Paillé P, (2016). Linking corporate policy and supervisory support with environmental citizenship behaviors: The role of employee environmental beliefs and commitment. Journal of Business Ethics, 137( 1): 129–148

[75] Reed M, (2008). Stakeholder participation for environmental management: A literature review. Biological Conservation, 141( 10): 2417–2431

[76] Roy R B, Rokonuzzaman Md, Amin N, Mishu M K, Alahakoon S, Rahman S, Mithulananthan N, Rahman K S, Shakeri M, Pasupuleti J, (2021). A Comparative Performance Analysis of ANN Algorithms for MPPT Energy Harvesting in Solar PV System. IEEE Access, 9: 102137–102152

[77] Smail E (2022). Australian Conservation Foundation Launches Legal Action Against Woodside. Climate News Australia. Available at the website of climatenewsaustralia.com/australian-conservation-foundation-launches-legal-action-against-woodside/

[78] Swalih M M, Ram R, Tew E, (2024). Environmental management accounting for strategic decision-making: A systematic literature review. Business Strategy and the Environment, 33( 7): 6335–6367

[79] Taylor K MRosca E (2024). Toward a moral approach to stakeholder management: Insights from the inclusion of marginalized stakeholders in the operations of social enterprises. International Journal of Operations & Production Management. 44 (10): 1831–1858

[80] Eemeren F H van Grootendorst R (2003). A Systematic Theory of Argumentation: The Pragma-dialectical Approach. Cambridge University Press

[81] Vayansky I, Kumar S A P, (2020). A review of topic modeling methods. Information Systems, 94: 101582

[82] Vining J, (1987). Environmental decisions: The interaction of emotions, information, and decision context. Journal of Environmental Psychology, 7( 1): 13–30

[83] Wallach H M (2006). Topic modeling: Beyond bag-of-words. In: Proceedings of the 23rd International Conference on Machine Learning, 977–984

[84] Xue J, Shen G Q, Li Y, Wang J, Zafar I, (2020). Dynamic stakeholder-associated topic modeling on public concerns in megainfrastructure projects: Case of Hong Kong–Zhuhai–Macao Bridge. Journal of Management in Engineering, 36( 6): 04020078

RIGHTS & PERMISSIONS

Higher Education Press