A5053418697- Sentiment Analysis and Opinion Mining
- Mobile Crowdsensing and Crowdsourcing
- Topic Modeling
- Personality Traits and Psychology
- Blind Source Separation Techniques
- Digital Marketing and Social Media
- Data Visualization and Analytics
- Consumer Behavior in Brand Consumption and Identification
- Open Source Software Innovations
- Advanced Algorithms and Applications
- Image Retrieval and Classification Techniques
- Natural Language Processing Techniques
- Spectroscopy and Chemometric Analyses
- Anomaly Detection Techniques and Applications
- AI in Service Interactions
- Advanced Text Analysis Techniques
- Neural Networks and Applications
- Image and Video Stabilization
- Ethics and Social Impacts of AI
- Adversarial Robustness in Machine Learning
- FinTech, Crowdfunding, Digital Finance
- Big Data and Business Intelligence
- Explainable Artificial Intelligence (XAI)
- Mental Health via Writing
- Digital Communication and Language
Nvidia (United States)
2023-2024
LinkedIn (United States)
2022
IBM Research - Almaden
2015-2021
University of Illinois Urbana-Champaign
2009-2014
Beijing Normal University
2006-2008
Institute of Electrical and Electronics Engineers
2007
University of Hong Kong
2007
Hong Kong University of Science and Technology
2007
Users are rapidly turning to social media request and receive customer service; however, a majority of these requests were not addressed timely or even at all. To overcome the problem, we create new conversational system automatically generate responses for users on media. Our is integrated with state-of-the-art deep learning techniques trained by nearly 1M Twitter conversations between agents from over 60 brands. The evaluation reveals that 40% emotional, about as good human in showing...
Hundreds of thousands crowdfunding campaigns have been launched, but more than half them failed. To better understand the factors affecting campaign outcomes, this paper targets content and usage patterns project updates -- communications intended to keep potential funders aware a campaign's progress. We analyzed large corpus on Kickstarter, one largest platforms. Using semantic analysis techniques, we derived taxonomy types created during campaigns, found discrepancies between design intent...
Feedback on designs is critical for helping users iterate toward effective solutions. This paper presents Voyant, a novel system giving access to non-expert crowd receive perception-oriented feedback their from selected audience. Based formative study, the generates elements seen in design, order which are noticed, impressions formed when design first viewed, and interpretation of relative guidelines domain user's stated goals. An evaluation was conducted with designs. Users reported about...
Chatbot has become an important solution to rapidly increasing customer care demands on social media in recent years. However, current work chatbot for ignores a key impact user experience - tones. In this work, we create novel tone-aware that generates toned responses requests media. We first conduct formative research, which the effects of tones are studied. Significant and various influences different uncovered study. With knowledge tones, design deep learning based takes tone information...
Social media platforms provide an enormous public repository of textual data from which valuable information can be extracted. We show that firms extract business intelligence social bearing on important application, measuring brand personality. Specifically, we develop a text analytics framework integrates different distinct sources generated by consumers, employees, and firms, to measure Based Elastic-Net regression analyses large corpus data, including self-descriptions 1,996,214...
The interpretability or explainability of AI systems (XAI) has been a topic gaining renewed attention in recent years across and HCI communities. Recent work drawn to the emergent requirements situ, applied projects, yet further exploratory is needed more fully understand this space. This paper investigates projects reports on qualitative interview study individuals working at large technology consulting company. Presenting an empirical understanding range stakeholders industrial also draws...
We present a comprehensive system for weather data visualization. Weather are multivariate and contain vector fields formed by wind speed direction. Several well-established visualization techniques such as parallel coordinates polar systems integrated into our system. also develop various novel methods, including circular pixel bar charts embedded systems, enhanced with S-shape axis, weighted complete graphs. Our was used to analyze the air pollution problem in Hong Kong some interesting...
Predicting personality is essential for social applications supporting human-centered activities, yet prior modeling methods with users’ written text require too much input data to be realistically used in the context of media. In this work, we aim drastically reduce requirement and develop a model that applicable most users on Twitter. Our integrates Word Embedding features Gaussian Processes regression. Based evaluation over 1.3K Twitter, find our achieves comparable or better accuracy...
Critique is an indispensible part of creative work and many online communities have formed for this shared purpose. As design choices within the can impact effectiveness critiques produced, it important to study these offer guidance decisions. In paper, we report results a case exploring one large community dedicated critique in domain digital photography. We analyzed corpus interaction data understand benefit participation, response dynamics, factors predicting ratings, patterns reciprocal...
Crowd feedback systems offer designers an emerging approach for improving their designs, but there is little empirical evidence of the benefit these systems. This paper reports results a study using crowd system to iterate on visual designs. Users in introductory design course created initial designs satisfying brief and received revised was used generate again. format enabled us detect changes between how related those changes. Further, we analyzed value by comparing it with expert...
Chatbot has become an important solution to rapidly increasing customer care demands on social media in recent years. However, current work chatbot for ignores a key impact user experience - tones. In this work, we create novel tone-aware that generates toned responses requests media. We first conduct formative research, which the effects of tones are studied. Significant and various influences different uncovered study. With knowledge tones, design deep learning based takes tone information...
Brand personality has been shown to affect a variety of user behaviors such as individual preferences and social interactions. Despite intensive research efforts in human assessment, little is known about brand its relationship with media. Leveraging the theory marketing, we analyze how associates contributing factors embodied Based on analysis over 10K survey responses large corpus media data from 219 brands, quantify relative importance driving personality. The model developed achieves...
Successful human interactions are based on becoming aware of other's emotion and making adaptations accordingly. However, understanding is a complex task that has generated countless debates among researchers over the past decades. The abstract nature highlights need for new data-driven approach can better describe compare across fine-grained emotional states. In this study, we propose Seemo, novel neural embedding framework, which allows us to map emotions into vector space representations....
With the recent advances in using data analytics to automatically infer one's personality traits from their social media data, users are facing a growing tension between use of technology aid self development workplace and privacy concerns such use. Given richness that can be derived today varied sensitivity revealing it is non-trivial task for configure settings sharing protecting data. Here we present design, development, evaluation an interactive visualization tool, VeilMe, which helps...
Public perceptions of a brand is critical to its performance. While social media has demonstrated huge potential shape public brands, existing tools are not intuitive and explanatory for domain users use as they fail provide comprehensive analysis framework brands. In this paper, we present SocialBrands, novel visual tool managers understand brands on media. Social-Brands leverages personality in marketing literature computing approaches compute the from three driving factors (user imagery,...
Enterprise chatbots, powered by generative AI, are emerging as key applications to enhance employee productivity. Retrieval Augmented Generation (RAG), Large Language Models (LLMs), and orchestration frameworks like Langchain Llamaindex crucial for building these chatbots. However, creating effective enterprise chatbots is challenging requires meticulous RAG pipeline engineering. This includes fine-tuning embeddings LLMs, extracting documents from vector databases, rephrasing queries,...
We propose a novel method, the complete two-dimensional principal component analysis (complete 2DPCA), for image features extraction. Compared to original 2DPCA, 2DPCA not only gain higher recognition rate, but also reduce feature coefficients needed face recognition. Complete is based on 2D matrices. Two covariance matrices are constructed directly using matrix and theirs eigenvectors derived Our experiments were performed ORL database, experimental results show that proposed method has an...
Idea pipelines enable open innovation within organizations but require the evaluation teams to assess large numbers of ideas. To help filter promising ideas, community voting is often included as part pipeline outcome rarely aligns with ideas selected by team. address this problem, we introduce a new scoring model for increasing findability idea pipelines. In model, each participant need only score subset are scored independently, and individual scores can be aggregated. We tested on an...
In this paper, we present computational models to predict Twitter users' attitude towards a specific brand through their personal and social characteristics. We also likelihood of taking different actions based on attitudes. order operationalize our research actions, collected ground-truth data surveys users. have conducted experiments using two real world datasets validate the effectiveness action prediction framework. Finally, show how can be integrated with visual analytics system for...
Personality traits have long been shown to contribute consumer behaviors. Besides the main effect of personality, it seems also plausible that other factors may impact this relationship. To understand such situational effects, in study we analyze two possible variables, namely, income and needs. We conduct extensive analysis on a large industry dataset across over 100 product categories. For each category, build prediction model for consumption decision based derived personality features....