A5062817658- Topic Modeling
- Natural Language Processing Techniques
- Explainable Artificial Intelligence (XAI)
- AI in Service Interactions
- Ethics and Social Impacts of AI
- Software Engineering Research
- Machine Learning and Data Classification
- Scientific Computing and Data Management
- Artificial Intelligence in Healthcare and Education
- Speech and dialogue systems
- Big Data and Business Intelligence
- Machine Learning in Healthcare
- Multimodal Machine Learning Applications
- Data Visualization and Analytics
- Innovative Human-Technology Interaction
- Machine Learning and Algorithms
- Social Robot Interaction and HRI
- Open Source Software Innovations
- Technology Use by Older Adults
- Team Dynamics and Performance
- Usability and User Interface Design
- Online Learning and Analytics
- Privacy, Security, and Data Protection
- Persona Design and Applications
- Innovative Teaching and Learning Methods
Northeastern University
2022-2025
Boston University
2022-2024
Hong Kong University of Science and Technology
2022-2024
University of Hong Kong
2022-2024
University of Science and Technology
2024
Universidad del Noreste
2022-2024
Cambridge Scientific (United States)
2020-2023
Rensselaer Polytechnic Institute
2022-2023
University of North Carolina Health Care
2023
University of North Carolina at Chapel Hill
2023
With the rise of big data, there has been an increasing need for practitioners in this space and opportunity researchers to understand their workflows design new tools improve it. Data science is often described as data-driven, comprising unambiguous data proceeding through regularized steps analysis. However, view focuses more on abstract processes, pipelines, workflows, less how workers engage with data. In paper, we build work other CSCW HCI describing ways that scientists, scholars,...
The rapid advancement of artificial intelligence (AI) is changing our lives in many ways. One application domain data science. New techniques automating the creation AI, known as AutoAI or AutoML, aim to automate work practices scientists. systems are capable autonomously ingesting and pre-processing data, engineering new features, creating scoring models based on a target objectives (e.g. accuracy run-time efficiency). Though not yet widely adopted, we interested understanding how will...
Today, the prominence of data science within organizations has given rise to teams workers collaborating on extracting insights from data, as opposed individual scientists working alone. However, we still lack a deep understanding how collaborate in practice. In this work, conducted an online survey with 183 participants who work various aspects science. We focused their reported interactions each other (e.g., managers engineers) and different tools Jupyter Notebook). found that are...
Human-Computer Integration (HInt) is an emerging paradigm in which computational and human systems are closely interwoven. Integrating computers with the body not new. however, we believe that rapid technological advancements, increasing real-world deployments, growing ethical societal implications, it critical to identify agenda for future research. We present a set of challenges HInt research, formulated over course five-day workshop consisting 29 experts who have designed, deployed...
Artificial Intelligent (AI) and Machine Learning (ML) algorithms are coming out of research labs into the real-world applications, recent has focused a lot on Human-AI Interaction (HAI) Explainable AI (XAI). However, is not same as Collaboration. Collaboration involves mutual goal understanding, preemptive task co-management shared progress tracking. Most human activities today done collaboratively, thus, to integrate already-complicated workflow, it critical bring Computer-Supported...
Artificial intelligence (AI) technology has been increasingly used in the implementation of advanced Clinical Decision Support Systems (CDSS). Research demonstrated potential usefulness AI-powered CDSS (AI-CDSS) clinical decision making scenarios. However, post-adoption user perception and experience remain understudied, especially developing countries. Through observations interviews with 22 clinicians from 6 rural clinics China, this paper reports various tensions between design an AI-CDSS...
The development of AI applications is a multidisciplinary effort, involving multiple roles collaborating with the developers, an umbrella term we use to include data scientists and other AI-adjacent on same team. During these collaborations, there knowledge mismatch between who are skilled in science, external stakeholders typically not. This difference leads communication gaps, onus falls developers explain science concepts their collaborators. In this paper, report study including analyses...
Despite its benefits for children's skill development and parent-child bonding, many parents do not often engage in interactive storytelling by having story-related dialogues with their child due to limited availability or challenges coming up appropriate questions. While recent advances made AI generation of questions from stories possible, the fully-automated approach excludes parent involvement, disregards educational goals, underoptimizes engagement. Informed need-finding interviews...
Advances in large language models (LLMs) have empowered a variety of applications. However, there is still significant gap research when it comes to understanding and enhancing the capabilities LLMs field mental health. In this work, we present comprehensive evaluation multiple on various health prediction tasks via online text data, including Alpaca, Alpaca-LoRA, FLAN-T5, GPT-3.5, GPT-4. We conduct broad range experiments, covering zero-shot prompting, few-shot instruction fine-tuning. The...
Artificial intelligence (AI)-driven chatbots are increasingly being used in health care, but most designed for a specific population and evaluated controlled settings. There is little research documenting how consumers (eg, patients caregivers) use self-diagnosis purposes real-world scenarios.The aim of this was to understand context, what issues barriers exist their usage, the user experience novel technology can be improved.We employed data-driven approach analyze system log widely...
We are interested in increasing the ability of groups to collaborate efficiently by leveraging new advances AI and Conversational Agent (CA) technology. Given longstanding debate on necessity embodiment for CAs, bringing them requires answering questions whether how providing a CA with face affects its interaction humans group. explored these comparing group decision-making sessions facilitated an embodied agent, versus voice-only agent. Results experiment 20 user revealed that while...
Many conversational agents (CAs) are developed to answer users' questions in a specialized domain. In everyday use of CAs, user experience may extend beyond satisfying information needs the enjoyment conversations with some which represent playful interactions. By studying field deployment Human Resource chatbot, we report on interest areas interactions inform development CAs. Through lens statistical modeling, also highlight rich signals for inferring satisfaction instrumental usage and...
Many approaches to extract multiple relations from a paragraph require passes over the paragraph. In practice, are computationally expensive and this makes difficult scale longer paragraphs larger text corpora. work, we focus on task of relation extractions by encoding only once. We build our solution upon pre-trained self-attentive models (Transformer), where first add structured prediction layer handle extraction between entity pairs, then enhance embedding capture relational information...
We explore trust in a relatively new area of data science: Automated Machine Learning (AutoML). In AutoML, AI methods are used to generate and optimize machine learning models by automatically engineering features, selecting models, optimizing hyperparameters. this paper, we seek understand what kinds information influence scientists' the produced AutoML? operationalize as willingness deploy model using automated methods. report results from three studies -- qualitative interviews,...
Results of radiology imaging studies are not typically comprehensible to patients. With the advances in artificial intelligence (AI) technology recent years, it is expected that AI can aid patients’ understanding data. The aim this study understand perceptions and acceptance using interpret their reports. We conducted semi-structured interviews with 13 participants elicit reflections pertaining use report interpretation. A thematic analysis approach was employed analyze interview...
Computational notebooks allow data scientists to express their ideas through a combination of code and documentation. However, often pay attention only the code, neglect creating or updating documentation during quick iterations. Inspired by human practices learned from 80 highly-voted Kaggle notebooks, we design implement Themisto, an automated generation system explore how human-centered AI systems can support in machine learning scenario. Themisto facilitates creation via three...
Advances in large language models (LLMs) have empowered a variety of applications. However, there is still significant gap research when it comes to understanding and enhancing the capabilities LLMs field mental health. In this work, we present first comprehensive evaluation multiple LLMs, including Alpaca, Alpaca-LoRA, FLAN-T5, GPT-3.5, GPT-4, on various health prediction tasks via online text data. We conduct broad range experiments, covering zero-shot prompting, few-shot instruction...
The widespread use of Large Language Model (LLM)-based conversational agents (CAs), especially in high-stakes domains, raises many privacy concerns. Building ethical LLM-based CAs that respect user requires an in-depth understanding the risks concern users most. However, existing research, primarily model-centered, does not provide insight into users' perspectives. To bridge this gap, we analyzed sensitive disclosures real-world ChatGPT conversations and conducted semi-structured interviews...
Despite the plethora of telehealth applications to assist home-based older adults and healthcare providers, basic messaging phone calls are still most common communication methods, which suffer from limited availability, information loss, process inefficiencies. One promising solution facilitate patient-provider is leverage large language models (LLMs) with their powerful natural conversation summarization capability. However, there a understanding LLMs' role during communication. We first...
In this paper, we introduce the design and evaluation of an LLM-based AI agent for human-agent interaction in Virtual Reality (VR). Our system leverages GPT-4, a Large Language Model (LLM) to simulate human behavior. agent, deployed VRChat as Non-playable Character (NPC), exhibits ability respond player by providing context-relevant responses followed appropriate facial expressions body gestures. preliminary yielded most optimal parameters generating plausible responses. With our system, lay...
Today's commercially available word processors allow people to write collaboratively in the cloud, both familiar asynchronous mode and now synchronous as well. This opens up new ways of working together. We examined data traces collaborative writing behavior student teams’ use Google Docs discover how they are together now. found that teams synchronously asynchronously, take fluid roles editing documents, show a variety styles writing, including from scratch, beginning with an outline,...
Collaborative writing is on the increase. In order to write well together, authors often need be aware of who has done what recently. We offer a new tool, DocuViz, that displays entire revision history Google Docs, showing more than one-step-at-a-time view now shown in and tracking changes Word. introduce tool present cases which potential useful: To themselves see recent "seismic activity," indicating where particular co-author might want pay attention, instructors contributed were made...
Recently Nonprofit organizations (NPOs) are adopting more and data-driven approaches to their work, yet NPOs often lack appropriate tools expertise in such data related works. To compensate, many using a new form of collaboration, civic hackathons, leverage on external volunteers' expertise. In this paper, we sought understand how hackathons could generate impactful analytics for NPOs' support collaborative during hackathons. We collected various types (observations, surveys, interviews)...