Global Evolution of Social Responsibility in Smart-Service Industries: Insights From a Cross-Sector Hybrid Large Language Models Approach

Ziyuan Xia, Saixing Zeng, Anchen Sun, Huabin Sun, Xiaodong Cai

Abstract The rapid development of smart services has driven the evolution of social responsibility among global enterprises, while also presenting new challenges to their operational management. In the continuous iterations of smart services, the emerging digital ecosystem has demonstrated multidimensional characteristics, virtual-real integration, and multi-stakeholder interactions in management practices. This study introduces smart service social responsibility (SSSR), a comprehensive framework that extends traditional CSR and ESG models by integrating multidimensional, cross-space, and multi-stakeholder panoramic analyses to address the unique ethical and operational challenges of the digital ecosystem. Using a hybrid text-analysis methodology (TF-IDF scoring and LLM-based evaluation), we analyze 7858 sustainability reports across five major sectors (consumer goods, technology, financial, healthcare, and services) to reveal how firms prioritize sustainability issues and identify sector-specific patterns in smart-service industries. Our analysis reveals that environmental topics dominate, accounting for an average of 49.0% of dimension-level mentions and leading in four of the five sectors studied, whereas legal and ethical themes receive 42.25% fewer mentions on average. Meanwhile, physical space topics constitute nearly three-quarters (76.5%) of the total, in contrast to virtual space themes, which represent approximately one-quarter (23.5%). Furthermore, analysis of stakeholder attention reveals a strong focus on platforms (42.4%) and communities (23.8%), which together account for over 66.2% of the discourse, while emerging agents, such as algorithm engineers and smart bots, remain significantly underrepresented. The novelty of our research is demonstrated through uncovering how firms prioritize topics of social responsibility in sustainability reporting and revealing sector-specific patterns that highlight prominently featured content. These insights offer important guidance for regulators, businesses, and investors seeking to align smart-service frontiers with responsible practices.

My role: I designed the SSSR framework and the hybrid TF-IDF + LLM methodology behind the analysis.

Who Said What (WSW 2.0)? Enhanced Automated Analysis of Preschool Classroom Speech

Anchen Sun, Tiantian Feng, Gabriela Gutierrez, Juan J Londono, Anfeng Xu, Batya Elbaum, Shrikanth Narayanan, Lynn K Perry, Daniel S Messinger

Abstract This paper introduces an automated framework (WSW 2.0) for analyzing vocal interactions in preschool classrooms, enhancing both accuracy and scalability through the integration of wav2vec2-based speaker classification and Whisper (large-v2 and large-v3) speech transcription. A total of 235 minutes of audio recordings (160 minutes from 12 children and 75 minutes from 5 teachers), were used to compare system outputs to expert human annotations. WSW 2.0 achieves a weighted F1 score of .845, accuracy of .846, and an error-corrected kappa of .672 for speaker classification (child vs. teacher). Transcription quality is moderate to high with word error rates of .119 for teachers and .238 for children. WSW 2.0 exhibits relatively high absolute agreement intraclass correlations (ICC) with expert transcriptions for a range of classroom language features. These include teacher and child mean utterance length, lexical diversity, question asking, and responses to questions and other utterances, which show absolute agreement intraclass correlations between .64 and .98. To establish scalability, we apply the framework to an extensive dataset spanning two years and over 1,592 hours of classroom audio recordings, demonstrating the framework’s robustness for broad real-world applications. These findings highlight the potential of deep learning and natural language processing techniques to revolutionize educational research by providing accurate measures of key features of preschool classroom speech, ultimately guiding more effective intervention strategies and supporting early childhood language development.

Assessing Sustainable Practices in Architecture: A Data-Driven Analysis of LEED Certification Adoption and Impact in Top Firms from 2000 to 2023

Jingyi Xu, Minghui Cheng, Anchen Sun

Abstract Amid increasing global environmental concerns, the architectural industry is under increasing pressure to implement sustainable practices. Leadership in Energy and Environmental Design (LEED) certification has become a crucial benchmark for assessing green building practices. This study investigates the adoption and impact of LEED-certified projects within leading architectural firms from 2000 to 2023, utilizing a novel data mining framework to scan extensive datasets on LEED projects and firm operations. We introduce two key metrics: the Weighted LEED Achieved Score (WLAS) and the Green Impact Ratio (GIR), which evaluate the sustainability efforts of firms in relation to their market size and project scale. These metrics yield insights into how firms incorporate sustainability into their business and the environmental outcomes of their projects. Our research uncovers significant trends in LEED standard adoption, illustrating a strengthening commitment to sustainable buildings. The analysis underscores the strategic importance of these practices for securing a competitive edge and aligning with global sustainability objectives. This paper contributes to the sustainable architecture discourse by providing fresh insights into the integration and effectiveness of LEED certification among top firms and offering a comprehensive framework for evaluating the environmental and economic aspects of sustainability in architecture.

Who Said What? An Automated Approach to Analyzing Speech in Preschool Classrooms

Anchen Sun, Juan J Londono, Batya Elbaum, Luis Estrada, Roberto Jose Lazo, Laura Vitale, Hugo Gonzalez Villasanti, Riccardo Fusaroli, Lynn K Perry, Daniel S Messinger

Abstract Young children spend substantial portions of their waking hours in noisy preschool classrooms. In these environments, children’s vocal interactions with teachers are critical contributors to their language outcomes, but manually transcribing these interactions is prohibitive. Using audio from child- and teacher-worn recorders, we propose an automated framework that uses open source software both to classify speakers (ALICE) and to transcribe their utterances (Whisper). We compare results from our framework to those from a human expert for 110 minutes of classroom recordings, including 85 minutes from child-word microphones (n=4 children) and 25 minutes from teacher-worn microphones (n=2 teachers). The overall proportion of agreement, that is, the proportion of correctly classified teacher and child utterances, was .76, with an error-corrected kappa of .50 and a weighted F1 of .76. The word error rate for both teacher and child transcriptions was .15, meaning that 15% of words would need to be deleted, added, or changed to equate the Whisper and expert transcriptions. Moreover, speech features such as the mean length of utterances in words, the proportion of teacher and child utterances that were questions, and the proportion of utterances that were responded to within 2.5 seconds were similar when calculated separately from expert and automated transcriptions. The results suggest substantial progress in analyzing classroom speech that may support children’s language development. Future research using natural language processing is under way to improve speaker classification and to analyze results from the application of the automated framework to a larger dataset containing classroom recordings from 13 children and 3 teachers observed on 17 occasions over one year.