Longqi Yang

Taiwei Shi, Zhuoer Wang, Longqi Yang et al. · amazon-science, microsoft-research

As large language models (LLMs) continue to advance, aligning these models with human preferences has emerged as a critical challenge. Traditional alignment methods, relying on human or LLM annotated datasets, are limited by their resource-intensive nature, inherent subjectivity, misalignment with real-world user preferences, and the risk of feedback loops that amplify model biases. To overcome these limitations, we introduce WildFeedback, a novel framework that leverages in-situ user feedback during conversations with LLMs to create preference datasets automatically. Given a corpus of multi-turn user-LLM conversation, WildFeedback identifies and classifies user feedback to LLM responses between conversation turns. The user feedback is then used to create examples of preferred and dispreferred responses according to users' preference. Our experiments demonstrate that LLMs fine-tuned on WildFeedback dataset exhibit significantly improved alignment with user preferences, as evidenced by both traditional benchmarks and our proposed checklist-guided evaluation. By incorporating in-situ feedback from actual users, WildFeedback addresses the scalability, subjectivity, and bias challenges that plague existing approaches, marking a significant step toward developing LLMs that are more responsive to the diverse and evolving needs of their users.

Chirag Shah, Ryen W. White, Reid Andersen et al.

Log data can reveal valuable information about how users interact with Web search services, what they want, and how satisfied they are. However, analyzing user intents in log data is not easy, especially for emerging forms of Web search such as AI-driven chat. To understand user intents from log data, we need a way to label them with meaningful categories that capture their diversity and dynamics. Existing methods rely on manual or machine-learned labeling, which are either expensive or inflexible for large and dynamic datasets. We propose a novel solution using large language models (LLMs), which can generate rich and relevant concepts, descriptions, and examples for user intents. However, using LLMs to generate a user intent taxonomy and apply it for log analysis can be problematic for two main reasons: (1) such a taxonomy is not externally validated; and (2) there may be an undesirable feedback loop. To address this, we propose a new methodology with human experts and assessors to verify the quality of the LLM-generated taxonomy. We also present an end-to-end pipeline that uses an LLM with human-in-the-loop to produce, refine, and apply labels for user intent analysis in log data. We demonstrate its effectiveness by uncovering new insights into user intents from search and chat logs from the Microsoft Bing commercial search engine. The proposed work's novelty stems from the method for generating purpose-driven user intent taxonomies with strong validation. This method not only helps remove methodological and practical bottlenecks from intent-focused research, but also provides a new framework for generating, validating, and applying other kinds of taxonomies in a scalable and adaptable way with reasonable human effort.

Sarkar Snigdha Sarathi Das, Chirag Shah, Mengting Wan et al.

The traditional Dialogue State Tracking (DST) problem aims to track user preferences and intents in user-agent conversations. While sufficient for task-oriented dialogue systems supporting narrow domain applications, the advent of Large Language Model (LLM)-based chat systems has introduced many real-world intricacies in open-domain dialogues. These intricacies manifest in the form of increased complexity in contextual interactions, extended dialogue sessions encompassing a diverse array of topics, and more frequent contextual shifts. To handle these intricacies arising from evolving LLM-based chat systems, we propose joint dialogue segmentation and state tracking per segment in open-domain dialogue systems. Assuming a zero-shot setting appropriate to a true open-domain dialogue system, we propose S3-DST, a structured prompting technique that harnesses Pre-Analytical Recollection, a novel grounding mechanism we designed for improving long context tracking. To demonstrate the efficacy of our proposed approach in joint segmentation and state tracking, we evaluate S3-DST on a proprietary anonymized open-domain dialogue dataset, as well as publicly available DST and segmentation datasets. Across all datasets and settings, S3-DST consistently outperforms the state-of-the-art, demonstrating its potency and robustness the next generation of LLM-based chat systems.

Sheshera Mysore, Zhuoran Lu, Mengting Wan et al.

Powerful large language models have facilitated the development of writing assistants that promise to significantly improve the quality and efficiency of composition and communication. However, a barrier to effective assistance is the lack of personalization in LLM outputs to the author's communication style, specialized knowledge, and values. In this paper, we address this challenge by proposing Pearl, a LLM writing assistant personalized with a retriever that is trained to be generation-calibrated for personalization. Generation calibration ensures that our retriever selects historic user authored documents to augment an LLM prompt such that they are likely to help an LLM generation better adhere to a users' preferences. We propose two key novelties for training such a retriever: (1) A training data selection method that identifies user requests likely to benefit from personalization and documents that provide that benefit; and (2) A scale-calibrating KL-divergence objective that ensures that our retriever scores remain proportional to the downstream generation quality from using the document for personalized generation. In a series of holistic evaluations, we demonstrate the effectiveness of Pearl in generating long-form texts on multiple social media datasets. Finally, we demonstrate how a generation-calibrated retriever can double as a performance predictor -- detecting low quality retrieval, and improving potentially under-performing outputs via revision with LLMs.

Bowen Jiang, Taiwei Shi, Ryo Kamoi et al.

This paper introduces OMAR: One Model, All Roles, a reinforcement learning framework that enables AI to develop social intelligence through multi-turn, multi-agent conversational self-play. Unlike traditional paradigms that rely on static, single-turn optimizations, OMAR allows a single model to role-play all participants in a conversation simultaneously, learning to achieve long-term goals and complex social norms directly from dynamic social interaction. To ensure training stability across long dialogues, we implement a hierarchical advantage estimation that calculates turn-level and token-level advantages. Evaluations in the SOTOPIA social environment and Werewolf strategy games show that our trained models develop fine-grained, emergent social intelligence, such as empathy, persuasion, and compromise seeking, demonstrating the effectiveness of learning collaboration even under competitive scenarios. While we identify practical challenges like reward hacking, our results show that rich social intelligence can emerge without human supervision. We hope this work incentivizes further research on AI social intelligence in group conversations.

Marios Papachristou, Longqi Yang, Chin-Chia Hsu

In various work contexts, such as meeting scheduling, collaborating, and project planning, collective decision-making is essential but often challenging due to diverse individual preferences, varying work focuses, and power dynamics among members. To address this, we propose a system leveraging Large Language Models (LLMs) to facilitate group decision-making by managing conversations and balancing preferences among individuals. Our system aims to extract individual preferences from each member's conversation with the system and suggest options that satisfy the preferences of the members. We specifically apply this system to corporate meeting scheduling. We create synthetic employee profiles and simulate conversations at scale, leveraging LLMs to evaluate the system performance as a novel approach to conducting a user study. Our results indicate efficient coordination with reduced interactions between the members and the LLM-based system. The system refines and improves its proposed options over time, ensuring that many of the members' individual preferences are satisfied in an equitable way. Finally, we conduct a survey study involving human participants to assess our system's ability to aggregate preferences and reasoning about them. Our findings show that the system exhibits strong performance in both dimensions.

Longqi Yang, Cheng-Kang Hsieh, Hongjian Yang et al.

Nutrient-based meal recommendations have the potential to help individuals prevent or manage conditions such as diabetes and obesity. However, learning people's food preferences and making recommendations that simultaneously appeal to their palate and satisfy nutritional expectations are challenging. Existing approaches either only learn high-level preferences or require a prolonged learning period. We propose Yum-me, a personalized nutrient-based meal recommender system designed to meet individuals' nutritional expectations, dietary restrictions, and fine-grained food preferences. Yum-me enables a simple and accurate food preference profiling procedure via a visual quiz-based user interface, and projects the learned profile into the domain of nutritionally appropriate food options to find ones that will appeal to the user. We present the design and implementation of Yum-me, and further describe and evaluate two innovative contributions. The first contriution is an open source state-of-the-art food image analysis model, named FoodDist. We demonstrate FoodDist's superior performance through careful benchmarking and discuss its applicability across a wide array of dietary applications. The second contribution is a novel online learning framework that learns food preference from item-wise and pairwise image comparisons. We evaluate the framework in a field study of 227 anonymous users and demonstrate that it outperforms other baselines by a significant margin. We further conducted an end-to-end validation of the feasibility and effectiveness of Yum-me through a 60-person user study, in which Yum-me improves the recommendation acceptance rate by 42.63%.

Mengting Wan, Tara Safavi, Sujay Kumar Jauhar et al.

Transforming unstructured text into structured and meaningful forms, organized by useful category labels, is a fundamental step in text mining for downstream analysis and application. However, most existing methods for producing label taxonomies and building text-based label classifiers still rely heavily on domain expertise and manual curation, making the process expensive and time-consuming. This is particularly challenging when the label space is under-specified and large-scale data annotations are unavailable. In this paper, we address these challenges with Large Language Models (LLMs), whose prompt-based interface facilitates the induction and use of large-scale pseudo labels. We propose TnT-LLM, a two-phase framework that employs LLMs to automate the process of end-to-end label generation and assignment with minimal human effort for any given use-case. In the first phase, we introduce a zero-shot, multi-stage reasoning approach which enables LLMs to produce and refine a label taxonomy iteratively. In the second phase, LLMs are used as data labelers that yield training samples so that lightweight supervised classifiers can be reliably built, deployed, and served at scale. We apply TnT-LLM to the analysis of user intent and conversational domain for Bing Copilot (formerly Bing Chat), an open-domain chat-based search engine. Extensive experiments using both human and automatic evaluation metrics demonstrate that TnT-LLM generates more accurate and relevant label taxonomies when compared against state-of-the-art baselines, and achieves a favorable balance between accuracy and efficiency for classification at scale. We also share our practical experiences and insights on the challenges and opportunities of using LLMs for large-scale text mining in real-world applications.

Rupak Sarkar, Bahareh Sarrafzadeh, Nirupama Chandrasekaran et al.

Human-LLM conversations are increasingly becoming more pervasive in peoples' professional and personal lives, yet many users still struggle to elicit helpful responses from LLM Chatbots. One of the reasons for this issue is users' lack of understanding in crafting effective prompts that accurately convey their information needs. Meanwhile, the existence of real-world conversational datasets on the one hand, and the text understanding faculties of LLMs on the other, present a unique opportunity to study this problem, and its potential solutions at scale. Thus, in this paper we present the first LLM-centric study of real human-AI chatbot conversations, focused on investigating aspects in which user queries fall short of expressing information needs, and the potential of using LLMs to rewrite suboptimal user prompts. Our findings demonstrate that rephrasing ineffective prompts can elicit better responses from a conversational system, while preserving the user's original intent. Notably, the performance of rewrites improves in longer conversations, where contextual inferences about user needs can be made more accurately. Additionally, we observe that LLMs often need to -- and inherently do -- make \emph{plausible} assumptions about a user's intentions and goals when interpreting prompts. Our findings largely hold true across conversational domains, user intents, and LLMs of varying sizes and families, indicating the promise of using prompt rewriting as a solution for better human-AI interactions.

Shuaiqi Wang, Vikas Raunak, Arturs Backurs et al.

Differentially private (DP) synthetic data generation is a promising technique for utilizing private datasets that otherwise cannot be exposed for model training or other analytics. While much research literature has focused on generating private unstructured text and image data, in enterprise settings, structured data (e.g., tabular) is more common, often including natural language fields or components. Existing synthetic data evaluation techniques (e.g., FID) struggle to capture the structural properties and correlations of such datasets. In this work, we propose Struct-Bench, a framework and benchmark for evaluating synthetic datasets derived from structured datasets that contain natural language data. The Struct-Bench framework requires users to provide a representation of their dataset structure as a Context-Free Grammar (CFG). Our benchmark comprises 5 real-world and 2 synthetically generated datasets, each annotated with CFGs. We show that these datasets demonstrably present a great challenge even for state-of-the-art DP synthetic data generation methods. Struct-Bench also includes reference implementations of different metrics and a leaderboard, thereby providing researchers a standardized evaluation platform to benchmark and investigate privacy-preserving synthetic data generation methods. Further, we also present a case study showing how to use Struct-Bench to improve the synthetic data quality of Private Evolution (PE) on structured data. The benchmark and the leaderboard have been publicly made available at https://struct-bench.github.io.

Siddharth Suri, Scott Counts, Leijie Wang et al.

Until recently, search engines were the predominant method for people to access online information. The recent emergence of large language models (LLMs) has given machines new capabilities such as the ability to generate new digital artifacts like text, images, code etc., resulting in a new tool, a generative search engine, which combines the capabilities of LLMs with a traditional search engine. Through the empirical analysis of Bing Copilot (Bing Chat), one of the first publicly available generative search engines, we analyze the types and complexity of tasks that people use Bing Copilot for compared to Bing Search. Findings indicate that people use the generative search engine for more knowledge work tasks that are higher in cognitive complexity than were commonly done with a traditional search engine.

Ying-Chun Lin, Jennifer Neville, Jack W. Stokes et al.

Accurate and interpretable user satisfaction estimation (USE) is critical for understanding, evaluating, and continuously improving conversational systems. Users express their satisfaction or dissatisfaction with diverse conversational patterns in both general-purpose (ChatGPT and Bing Copilot) and task-oriented (customer service chatbot) conversational systems. Existing approaches based on featurized ML models or text embeddings fall short in extracting generalizable patterns and are hard to interpret. In this work, we show that LLMs can extract interpretable signals of user satisfaction from their natural language utterances more effectively than embedding-based approaches. Moreover, an LLM can be tailored for USE via an iterative prompting framework using supervision from labeled examples. The resulting method, Supervised Prompting for User satisfaction Rubrics (SPUR), not only has higher accuracy but is more interpretable as it scores user satisfaction via learned rubrics with a detailed breakdown.

Jing Ma, Ruocheng Guo, Mengting Wan et al.

Fair machine learning aims to mitigate the biases of model predictions against certain subpopulations regarding sensitive attributes such as race and gender. Among the many existing fairness notions, counterfactual fairness measures the model fairness from a causal perspective by comparing the predictions of each individual from the original data and the counterfactuals. In counterfactuals, the sensitive attribute values of this individual had been modified. Recently, a few works extend counterfactual fairness to graph data, but most of them neglect the following facts that can lead to biases: 1) the sensitive attributes of each node's neighbors may causally affect the prediction w.r.t. this node; 2) the sensitive attributes may causally affect other features and the graph structure. To tackle these issues, in this paper, we propose a novel fairness notion - graph counterfactual fairness, which considers the biases led by the above facts. To learn node representations towards graph counterfactual fairness, we propose a novel framework based on counterfactual data augmentation. In this framework, we generate counterfactuals corresponding to perturbations on each node's and their neighbors' sensitive attributes. Then we enforce fairness by minimizing the discrepancy between the representations learned from the original graph and the counterfactuals for each node. Experiments on both synthetic and real-world graphs show that our framework outperforms the state-of-the-art baselines in graph counterfactual fairness, and also achieves comparable prediction performance.

Rosie Jones, Hamed Zamani, Markus Schedl et al.

Podcasts are spoken documents across a wide-range of genres and styles, with growing listenership across the world, and a rapidly lowering barrier to entry for both listeners and creators. The great strides in search and recommendation in research and industry have yet to see impact in the podcast space, where recommendations are still largely driven by word of mouth. In this perspective paper, we highlight the many differences between podcasts and other media, and discuss our perspective on challenges and future research directions in the domain of podcast information access.

Jenna Butler, Mary Czerwinski, Shamsi Iqbal et al.

We now turn to understanding the impact that COVID-19 had on the personal productivity and well-being of information workers as their work practices were impacted by remote work. This chapter overviews people's productivity, satisfaction, and work patterns, and shows that the challenges and benefits of remote work are closely linked. Looking forward, the infrastructure surrounding work will need to evolve to help people adapt to the challenges of remote and hybrid work.

Longqi Yang, Sonia Jaffe, David Holtz et al.

The COVID-19 pandemic has had a wide-ranging impact on information workers such as higher stress levels, increased workloads, new workstreams, and more caregiving responsibilities during lockdown. COVID-19 also caused the overwhelming majority of information workers to rapidly shift to working from home (WFH). The central question this work addresses is: can we isolate the effects of WFH on information workers' collaboration activities from all other factors, especially the other effects of COVID-19? This is important because in the future, WFH will likely to be more common than it was prior to the pandemic. We use difference-in-differences (DiD), a causal identification strategy commonly used in the social sciences, to control for unobserved confounding factors and estimate the causal effect of WFH. Our analysis relies on measuring the difference in changes between those who WFH prior to COVID-19 and those who did not. Our preliminary results suggest that on average, people spent more time on collaboration in April (Post WFH mandate) than in February (Pre WFH mandate), but this is primarily due to factors other than WFH, such as lockdowns during the pandemic. The change attributable to WFH specifically is in the opposite direction: less time on collaboration and more focus time. This reversal shows the importance of using causal inference: a simple analysis would have resulted in the wrong conclusion. We further find that the effect of WFH is moderated by individual remote collaboration experience prior to WFH. Meanwhile, the medium for collaboration has also shifted due to WFH: instant messages were used more, whereas scheduled meetings were used less. We discuss design implications -- how future WFH may affect focused work, collaborative work, and creative work.

Longqi Yang, Chen Fang, Hailin Jin et al.

Complex design tasks often require performing diverse actions in a specific order. To (semi-)autonomously accomplish these tasks, applications need to understand and learn a wide range of design procedures, i.e., Creative Procedural-Knowledge (CPK). Prior knowledge base construction and mining have not typically addressed the creative fields, such as design and arts. In this paper, we formalize an ontology of CPK using five components: goal, workflow, action, command and usage; and extract components' values from online design tutorials. We scraped 19.6K tutorial-related webpages and built a web application for professional designers to identify and summarize CPK components. The annotated dataset consists of 819 unique commands, 47,491 actions, and 2,022 workflows and goals. Based on this dataset, we propose a general CPK extraction pipeline and demonstrate that existing text classification and sequence-to-sequence models are limited in identifying, predicting and summarizing complex operations described in heterogeneous styles. Through quantitative and qualitative error analysis, we discuss CPK extraction challenges that need to be addressed by future research.

Longqi Yang, Diana Freed, Alex Wu et al.

Healthcare professionals use Activities of Daily Living (ADL) to characterize a patient's functional status and to evaluate the effectiveness of treatment plans. ADLs are traditionally measured using standardized text-based questionnaires and the only form of personalization is in the form of question branching logic. Pervasive smartphone adoption makes it feasible to consider more frequent patient-reporting on ADLs. However, asking generic sets of questions repeatedly introduces user burden and fatigue that threatens to interfere with their utility. We introduce an approach called YADL (Your Activities of Daily Living) which uses images of ADLs and personalization to improve survey efficiency and the patient-experience. It offers several potential benefits: wider coverage of ADLs, improved engagement, and accurate capture of individual health situations. In this paper, we discuss our system design and the wide applicability of the design process for survey tools in healthcare and beyond. Interactions with with a small number of patients with Arthritis throughout the design process have been promising and we share detailed insights.

Longqi Yang, Cheng-Kang Hsieh, Deborah Estrin

User preference profiling is an important task in modern online social networks (OSN). With the proliferation of image-centric social platforms, such as Pinterest, visual contents have become one of the most informative data streams for understanding user preferences. Traditional approaches usually treat visual content analysis as a general classification problem where one or more labels are assigned to each image. Although such an approach simplifies the process of image analysis, it misses the rich context and visual cues that play an important role in people's perception of images. In this paper, we explore the possibilities of learning a user's latent visual preferences directly from image contents. We propose a distance metric learning method based on Deep Convolutional Neural Networks (CNN) to directly extract similarity information from visual contents and use the derived distance metric to mine individual users' fine-grained visual preferences. Through our preliminary experiments using data from 5,790 Pinterest users, we show that even for the images within the same category, each user possesses distinct and individually-identifiable visual preferences that are consistent over their lifetime. Our results underscore the untapped potential of finer-grained visual preference profiling in understanding users' preferences.