Fact verification is a critical yet underexplored component of non-litigation legal practice. While existing research has examined automation in legal workflow and human-AI collaboration in high-stakes domains, little is known about how GenAI can support fact verification, a task that demands prudent judgment and strict accountability. To address this, we conducted semi-structured interviews with 18 lawyers to understand their current verification practices, attitudes toward GenAI adoption, and expectations for future systems. We found that while lawyers use GenAI for low-risk tasks like drafting and language optimization, concerns over accuracy, confidentiality, and liability are currently limiting its adoption for fact verification. These concerns translate into core design requirements for AI systems that are trustworthy and accountable. Based on these, we contribute design insights for human-AI collaboration in legal fact verification, emphasizing the development of auditable systems that balance efficiency with professional judgment and uphold ethical and legal accountability in high-stakes practice.

Senior architects are pivotal in shaping architectural projects, yet integrating Generative AI (GenAI) into their workflows presents notable challenges. A formative study (N=11) identified key pain points in their early concept design process. To address these, we developed EarlyArchi, a GenAI-driven system supporting automated concept generation and evaluation. In a within-subject study (N=13), participants used EarlyArchi for early-stage design tasks. Results showed enhanced perceived creativity, improved design competency, and more efficient ideation. However, concerns emerged regarding controllability and domain-specific accuracy, highlighting the need for features that preserve professional autonomy and trust. Further analysis revealed three GenAI involvement modes—fully AI-driven, GenAI-led, and human-led—emphasizing the importance of adaptive role allocation in balancing creative exploration with expert leadership. These findings offer insights into supporting senior architects through GenAI while identifying key considerations for designing future human–AI co-creation systems.

Starting the job hunt is often challenging for new graduates, who face barriers in translating experiences into actionable career profiles due to limited self-awareness and unclear skill mapping. Through formative study with new graduates and early-career professionals, we concluded specific challenges in experience extraction, skill organization, and expressive confidence. Drawing on these insights, we designed CareerCraft, an interactive system that scaffolds the construction of coherent career stories and supports tailored job searching via experience card extraction, guided profile building, and LLM-powered recommendations. In a within-subject evaluation (N=16), participants rated the efficacy of CareerCraft against the baseline condition without the tool in improving profile structuring, clarifying their self-awareness and competencies, and supporting informed job direction choices. Based on the findings, we concluded that CareerCraft offered a promising pathway to career readiness among new graduates to the workforce. We further summarized the design considerations for LLM products emphasizing on users’ self-exploration.

In architecture design, while Generative AI can effortlessly create initial prototypes, architects struggle to update designs to stakeholders’ evolving requirements. Through a formative study (N = 12), we identified specific obstacles that architecture designers face when meeting the dynamic design demands of various project stakeholders. We therefore developed ArchiConnect, a proof-of-concept interactive system that helps architects communicate with multiple stakeholders and update final design deliverables. ArchiConnect supports creativity and engagement by visualizing evolving demands, conflicts, and concept extractions from diverse stakeholders. We evaluated our system in a week-long user study (N = 8) with a simulated project. Participants found ArchiConnect effective for improving multi-stakeholder communication and management, describing it as intuitive and useful. Our findings offer design considerations for future AI tools to better handle dynamic stakeholder needs, including how to address sustainability requirements in line with development goals.

Junyu Luo, Zhizhuo Kou, Liming Yang, Xiao Luo, Jinsheng Huang, Zhiping Xiao, Jingshu Peng, Chengzhong Liu, Jiaming Ji, Xuanzhe Liu, Sirui Han, Ming Zhang, Yike Guo. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Chuxue Cao, Mengze Li, Juntao Dai, Jinluan Yang, Zijian Zhao, Shengyu Zhang, Weijie Shi, Chengzhong Liu, Sirui Han, Yike Guo. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

The escalating frequency and intensity of heat-related climate events, particularly heatwaves, emphasize the pressing need for advanced heat risk management strategies. Current approaches, primarily relying on numerical models, face challenges in spatial-temporal resolution and in capturing the dynamic interplay of environmental, social, and behavioral factors affecting heat risks. This has led to difficulties in translating risk assessments into effective mitigation actions. Recognizing these problems, we introduce a novel approach leveraging the burgeoning capabilities of Large Language Models (LLMs) to extract rich and contextual insights from news reports. We hence propose an LLM-empowered visual analytics system, Havior, that integrates the precise, data-driven insights of numerical models with nuanced news report information. This hybrid approach enables a more comprehensive assessment of heat risks and better identification, assessment, and mitigation of heat-related threats. The system incorporates novel visualization designs, such as "thermoglyph" and news glyph, enhancing intuitive understanding and analysis of heat risks. The integration of LLM-based techniques also enables advanced information retrieval and semantic knowledge extraction that can be guided by experts' analytics needs. We conducted an experiment on information extraction, a case study on the 2022 China Heatwave, and an expert survey & interview collaborated with six domain experts, demonstrating the usefulness of our system in providing in-depth and actionable insights for heat risk management.

Choice of color is critical to creating effective charts with an engaging, enjoyable, and informative reading experience. However, designing a good color palette for a chart is a challenging task for novice users who lack related design expertise. For example, they often find it difficult to articulate their abstract intentions and translate these intentions into effective editing actions to achieve a desired outcome. In this work, we present NL2Color, a tool that allows novice users to refine chart color palettes using natural language expressions of their desired outcomes. We first collected and categorized a dataset of 131 triplets, each consisting of an original color palette of a chart, an editing intent, and a new color palette designed by human experts according to the intent. Our tool employs a large language model (LLM) to substitute the colors in original palettes and produce new color palettes by selecting some of the triplets as few-shot prompts. To evaluate our tool, we conducted a comprehensive two-stage evaluation, including a crowd-sourcing study ( N=71) and a within-subjects user study ( N=12). The results indicate that the quality of the color palettes revised by NL2Color has no significantly large difference from those designed by human experts. The participants who used NL2Color obtained revised color palettes to their satisfaction in a shorter period and with less effort.

In Community-Based Question Answering (CQA) platforms, people can participate in discussions about non-factoid topics by marking their stances, providing premises, or arguing for the opinions they support, which forms “collective arguments”. The sustainable development of collective arguments relies on a big contributor base, yet most of the frequent CQA users are lurkers who seldom speak out. With a formative study, we identified detailed obstacles preventing lurkers from contributing to collective arguments. We consequently designed a processing pipeline for extracting and summarizing augmentative elements from question threads. Based on this we built CoArgue, a tool with navigation and chatbot features to support CQA lurkers’ motivation and ability in making contributions. Through a within-subject study (N=24), we found that, compared to a Quora-like baseline, participants perceived CoArgue as significantly more useful in enhancing their motivation and ability to join collective arguments and found the experience to be more engaging and productive.

Online fandom communities (OFCs) provide a convenient space for fans to create, collect, and discuss the content of their mutual interest (e.g., music artists). Real-world events could frequently attract outsiders to join OFCs, providing both the opportunity to expand the fan base and challenges to manage the community. However, it is unclear that how influxes of newcomers would influence the development of OFCs and what user behaviors may be correlated with their future engagement. To fill this gap, we took the music OFCs as the focus, and quantitatively analyzed user behaviors and their correlations with users' future engagement in the community. Results suggested that 1) event-induced newcomers expressed more hate speech and negative sentiment, praised less celebrity-related content (e.g., song, album), and interacted with narrower cohorts than existing members; 2) Although existing members tended to receive more upvotes during the events than before and after the events, newcomers showed an opposite trend; 3) keeping users' activeness, expressing positive sentiments, and having diverse interactions during periods of influx were helpful when maintaining members' future levels of engagement. This work deepened the understanding of fan behaviors in the dynamic period, and we discussed how our insights could benefit OFCs.

An effective task-oriented chatbot should be able to exert a certain level of Social Intelligence (SI), the ability to emulate human social behaviors to reduce user frustration and dissatisfaction. However, few studies explored using humor, a common rhetorical device in human-human interactions, to improve chatbots’ overall SI. To fill this gap, we proposed to apply self-mockery humor to a customer service chatbot in different interaction stages with users. We proposed a pipeline to create situated self-mockery for the chatbot and conducted a within-subject experiment (N=28) to compare it with a chatbot without self-mockery utterance. Results showed that the self-mockery chatbot was perceived as significantly funnier, more satisfactory, and delivering higher performance in two out of the five measured characteristics of SI with comparable performance in the rest. We further discussed how participants’ individual factors might affect the perceived helpfulness of self-mockery on SI and concluded with design considerations.

Generating educational questions of fairytales or storybooks is vital for improving children's literacy ability.However, it is challenging to generate questions that capture the interesting aspects of a fairytale story with educational meaningfulness.In this paper, we propose a novel question generation method that first learns the question type distribution of an input story paragraph, and then summarizes salient events which can be used to generate high-cognitive-demand questions.To train the event-centric summarizer, we finetune a pre-trained transformer-based sequenceto-sequence model using silver samples composed by educational question-answer pairs.On a newly proposed educational questionanswering dataset FairytaleQA, we show good performance of our method on both automatic and human evaluation metrics.Our work indicates the necessity of decomposing question type distribution learning and event-centric summary generation for educational question generation.

Community-based Question Answering (CQA) platforms can provide rich experience and suggestions for people who seek to construct Activity Plans (AP), such as bodybuilding or sightseeing. However, answer posts in CQA platforms could be too unstructured and overwhelming to be easily applied to AP construction, as validated by our formative study for understanding relevant user challenges. We therefore proposed an answer-post processing pipeline, based on which we built PlanHelper, a tool assisting users in processing the CQA information and constructing AP interactively. We conducted a within-subject study (N=24) with a Quora-like interface as the baseline. Results suggested that when creating AP with PlanHelper, users were significantly more satisfied with the informational support and more engaged during the interaction. Moreover, we performed an in-depth analysis on the user behaviors with PlanHelper and summarized the design considerations for such supporting tools.

Previous user experience research emphasizes meaning in interaction design beyond conventional interactive gestures. However, existing exemplars that successfully reify abstract meanings through interactions are usually case-specific, and it is currently unclear how to systematically create or extend meanings for general gesture-based interactions. We present Metaphoraction, a creativity support tool that formulates design ideas for gesture-based interactions to show metaphorical meanings with four interconnected components: gesture , action , object , and meaning . To represent the interaction design ideas with these four components, Metaphoraction links interactive gestures to actions based on the similarity of appearances, movements, and experiences; relates actions to objects by applying the immediate association; bridges objects and meanings by leveraging the metaphor TARGET-SOURCE mappings. We build a dataset containing 588,770 unique design idea candidates through surveying related research and conducting two crowdsourced studies to support meaningful gesture-based interaction design ideation. Five design experts validate that Metaphoraction can effectively support creativity and productivity during the ideation process. The paper concludes by presenting insights into meaningful gesture-based interaction design and discussing potential future uses of the tool.