Zhejiang University
AI Insights
- The proposed approach demonstrates the importance of rich contextual and behavioral signals for personalization, highlighting the need for more nuanced understanding of user characteristics and behaviors. [3]
- Previous work in user profile and recommender systems emphasizes the importance of rich contextual and behavioral signals for personalization. [3]
- Personalization in Conversational AI The paper presents a new approach to making conversational AI more personalized. [3]
- It focuses on understanding users' preferences and adapting to their needs, rather than relying solely on static information. [3]
- The paper presents a comprehensive approach to personalization in conversational AI, focusing on aligning large language models (LLMs) with user preferences. [2]
Abstract
The deployment of Large Language Models (LLMs) in interactive systems necessitates a deep alignment with the nuanced and dynamic preferences of individual users. Current alignment techniques predominantly address universal human values or static, single-turn preferences, thereby failing to address the critical needs of long-term personalization and the initial user cold-start problem. To bridge this gap, we propose PersonalAgent, a novel user-centric lifelong agent designed to continuously infer and adapt to user preferences. PersonalAgent constructs and dynamically refines a unified user profile by decomposing dialogues into single-turn interactions, framing preference inference as a sequential decision-making task. Experiments show that PersonalAgent achieves superior performance over strong prompt-based and policy optimization baselines, not only in idealized but also in noisy conversational contexts, while preserving cross-session preference consistency. Furthermore, human evaluation confirms that PersonalAgent excels at capturing user preferences naturally and coherently. Our findings underscore the importance of lifelong personalization for developing more inclusive and adaptive conversational agents. Our code is available here.
The deployment of Large Language Models (LLMs) in interactive systems necessitates a deep alignment with the nuanced and dynamic preferences of individual users. Current alignment techniques predominantly address universal human values or static, single-turn preferences, thereby failing to address the critical needs of long-term personalization and the initial user cold-start problem. To bridge this gap, we propose PersonalAgent, a novel user-centric lifelong agent designed to continuously infer and adapt to user preferences. PersonalAgent constructs and dynamically refines a unified user profile by decomposing dialogues into single-turn interactions, framing preference inference as a sequential decision-making task. Experiments show that PersonalAgent achieves superior performance over strong prompt-based and policy optimization baselines, not only in idealized but also in noisy conversational contexts, while preserving cross-session preference consistency. Furthermore, human evaluation confirms that PersonalAgent excels at capturing user preferences naturally and coherently. Our findings underscore the importance of lifelong personalization for developing more inclusive and adaptive conversational agents. Our code is available here.
Why we are recommending this paper?
Due to your Interest in: Personalization
This paper directly addresses personalization, a key interest, by exploring how LLMs can adapt to individual user preferences in dynamic dialogues. Given the focus on proactive personalization, this work aligns strongly with the user's desire for improved customer experiences.
Due to your Interest in: Personalization
This paper directly addresses personalization, a key interest, by exploring how LLMs can adapt to individual user preferences in dynamic dialogues. Given the focus on proactive personalization, this work aligns strongly with the user's desire for improved customer experiences.
UC Santa Barbara
AI Insights
- The study examines the performance of various language models on a dataset with diverse user profiles, highlighting the importance of personality traits and conversation style in determining likability. [3]
- The results show that top-performing models (GPT-5 and Claude Sonnet 4) are stable across different user types, while others struggle to adapt. [3]
- The introduction of Dynamic User Profile (DUP) improves performance for top models without additional training, indicating the potential of lightweight preference tracking. [3]
- Likability: The degree to which a language model is perceived as likable or relatable by users. [3]
- Conversation style preferences: The way individuals prefer to communicate, including aspects such as directness, formality, and conversation length. [3]
- Lightweight preference tracking (DUP) shows promise in improving performance without additional training. [3]
- The study's taxonomy of personality traits and conversation style preferences provides a comprehensive framework for understanding user behavior. [2]
Abstract
A personalized LLM should remember user facts, apply them correctly, and adapt over time to provide responses that the user prefers. Existing LLM personalization benchmarks are largely centered on two axes: accurately recalling user information and accurately applying remembered information in downstream tasks. We argue that a third axis, likability, is both subjective and central to user experience, yet under-measured by current benchmarks. To measure likability holistically, we introduce LikeBench, a multi-session, dynamic evaluation framework that measures likability across multiple dimensions by how much an LLM can adapt over time to a user's preferences to provide more likable responses. In LikeBench, the LLMs engage in conversation with a simulated user and learn preferences only from the ongoing dialogue. As the interaction unfolds, models try to adapt to responses, and after each turn, they are evaluated for likability across seven dimensions by the same simulated user. To the best of our knowledge, we are the first to decompose likability into multiple diagnostic metrics: emotional adaptation, formality matching, knowledge adaptation, reference understanding, conversation length fit, humor fit, and callback, which makes it easier to pinpoint where a model falls short. To make the simulated user more realistic and discriminative, LikeBench uses fine-grained, psychologically grounded descriptive personas rather than the coarse high/low trait rating based personas used in prior work. Our benchmark shows that strong memory performance does not guarantee high likability: DeepSeek R1, with lower memory accuracy (86%, 17 facts/profile), outperformed Qwen3 by 28% on likability score despite Qwen3's higher memory accuracy (93%, 43 facts/profile). Even SOTA models like GPT-5 adapt well in short exchanges but show only limited robustness in longer, noisier interactions.
A personalized LLM should remember user facts, apply them correctly, and adapt over time to provide responses that the user prefers. Existing LLM personalization benchmarks are largely centered on two axes: accurately recalling user information and accurately applying remembered information in downstream tasks. We argue that a third axis, likability, is both subjective and central to user experience, yet under-measured by current benchmarks. To measure likability holistically, we introduce LikeBench, a multi-session, dynamic evaluation framework that measures likability across multiple dimensions by how much an LLM can adapt over time to a user's preferences to provide more likable responses. In LikeBench, the LLMs engage in conversation with a simulated user and learn preferences only from the ongoing dialogue. As the interaction unfolds, models try to adapt to responses, and after each turn, they are evaluated for likability across seven dimensions by the same simulated user. To the best of our knowledge, we are the first to decompose likability into multiple diagnostic metrics: emotional adaptation, formality matching, knowledge adaptation, reference understanding, conversation length fit, humor fit, and callback, which makes it easier to pinpoint where a model falls short. To make the simulated user more realistic and discriminative, LikeBench uses fine-grained, psychologically grounded descriptive personas rather than the coarse high/low trait rating based personas used in prior work. Our benchmark shows that strong memory performance does not guarantee high likability: DeepSeek R1, with lower memory accuracy (86%, 17 facts/profile), outperformed Qwen3 by 28% on likability score despite Qwen3's higher memory accuracy (93%, 43 facts/profile). Even SOTA models like GPT-5 adapt well in short exchanges but show only limited robustness in longer, noisier interactions.
Why we are recommending this paper?
Due to your Interest in: Personalization
This research investigates LLM personalization through the lens of user 'likability,' a crucial factor in effective CRM and personalization strategies. The focus on remembering user facts and adapting responses is highly relevant to the user’s interest in optimizing personalization techniques.
Due to your Interest in: Personalization
This research investigates LLM personalization through the lens of user 'likability,' a crucial factor in effective CRM and personalization strategies. The focus on remembering user facts and adapting responses is highly relevant to the user’s interest in optimizing personalization techniques.
eBay Inc
AI Insights
- This paper presents a systematic comparison and analysis of various contextual features to refine search ranking on e-commerce platforms. [3]
- Mean Reciprocal Rank (MRR) is a metric used to evaluate the performance of search ranking algorithms, with higher values indicating better performance. [3]
- The study demonstrates significant improvements in MRR for sale items using various contextual features, underscoring their effectiveness in capturing user preferences. [3]
- Future work could focus on leveraging diverse interaction types beyond buyer clicks for richer contextualization. [3]
- The study relies heavily on existing literature and does not provide novel contributions or insights beyond the state-of-the-art. [3]
- The evaluation metrics used are limited to MRR, which may not capture other important aspects of search ranking performance. [3]
- Contextualization has been shown to be effective in improving search ranking performance in various studies. [3]
- This paper presents a systematic comparison and analysis of various contextual features to refine search ranking on e-commerce platforms, demonstrating significant improvements in Mean Reciprocal Rank (MRR) for sale items using these approaches. [3]
- Contextualization refers to the process of incorporating user behavior, preferences, and other relevant information into the search ranking algorithm to provide more personalized results. [2]
Abstract
In e-commerce shopping, aligning search results with a buyer's immediate needs and preferences presents a significant challenge, particularly in adapting search results throughout the buyer's shopping journey as they move from the initial stages of browsing to making a purchase decision or shift from one intent to another. This study presents a systematic approach to adapting e-commerce search results based on the current context. We start with basic methods and incrementally incorporate more contextual information and state-of-the-art techniques to improve the search outcomes. By applying this evolving contextual framework to items displayed on the search engine results page (SERP), we progressively align search outcomes more closely with the buyer's interests and current search intentions. Our findings demonstrate that this incremental enhancement, from simple heuristic autoregressive features to advanced sequence models, significantly improves ranker performance. The integration of contextual techniques enhances the performance of our production ranker, leading to improved search results in both offline and online A/B testing in terms of Mean Reciprocal Rank (MRR). Overall, the paper details iterative methodologies and their substantial contributions to search result contextualization on e-commerce platforms.
In e-commerce shopping, aligning search results with a buyer's immediate needs and preferences presents a significant challenge, particularly in adapting search results throughout the buyer's shopping journey as they move from the initial stages of browsing to making a purchase decision or shift from one intent to another. This study presents a systematic approach to adapting e-commerce search results based on the current context. We start with basic methods and incrementally incorporate more contextual information and state-of-the-art techniques to improve the search outcomes. By applying this evolving contextual framework to items displayed on the search engine results page (SERP), we progressively align search outcomes more closely with the buyer's interests and current search intentions. Our findings demonstrate that this incremental enhancement, from simple heuristic autoregressive features to advanced sequence models, significantly improves ranker performance. The integration of contextual techniques enhances the performance of our production ranker, leading to improved search results in both offline and online A/B testing in terms of Mean Reciprocal Rank (MRR). Overall, the paper details iterative methodologies and their substantial contributions to search result contextualization on e-commerce platforms.
Why we are recommending this paper?
Due to your Interest in: Paid Search
This paper tackles the core interest in paid search and e-commerce, specifically focusing on refining search results based on a buyer’s immediate actions. The focus on real-time adaptation aligns directly with the user’s interest in optimizing search ranking within a shopping context.
Due to your Interest in: Paid Search
This paper tackles the core interest in paid search and e-commerce, specifically focusing on refining search results based on a buyer’s immediate actions. The focus on real-time adaptation aligns directly with the user’s interest in optimizing search ranking within a shopping context.
Peking University
AI Insights
- It organizes operators along multiple orthogonal categorization dimensions, including modality, core vs. [3]
- PyTorch: An open-source machine learning library for Python. [3]
- domain-specific, and functional categories. [2]
- DataFlow is a unified data preparation framework that supports end-to-end LLM data preparation workflows. [1]
Abstract
The rapidly growing demand for high-quality data in Large Language Models (LLMs) has intensified the need for scalable, reliable, and semantically rich data preparation pipelines. However, current practices remain dominated by ad-hoc scripts and loosely specified workflows, which lack principled abstractions, hinder reproducibility, and offer limited support for model-in-the-loop data generation. To address these challenges, we present DataFlow, a unified and extensible LLM-driven data preparation framework. DataFlow is designed with system-level abstractions that enable modular, reusable, and composable data transformations, and provides a PyTorch-style pipeline construction API for building debuggable and optimizable dataflows. The framework consists of nearly 200 reusable operators and six domain-general pipelines spanning text, mathematical reasoning, code, Text-to-SQL, agentic RAG, and large-scale knowledge extraction. To further improve usability, we introduce DataFlow-Agent, which automatically translates natural-language specifications into executable pipelines via operator synthesis, pipeline planning, and iterative verification. Across six representative use cases, DataFlow consistently improves downstream LLM performance. Our math, code, and text pipelines outperform curated human datasets and specialized synthetic baselines, achieving up to +3\% execution accuracy in Text-to-SQL over SynSQL, +7\% average improvements on code benchmarks, and 1--3 point gains on MATH, GSM8K, and AIME. Moreover, a unified 10K-sample dataset produced by DataFlow enables base models to surpass counterparts trained on 1M Infinity-Instruct data. These results demonstrate that DataFlow provides a practical and high-performance substrate for reliable, reproducible, and scalable LLM data preparation, and establishes a system-level foundation for future data-centric AI development.
The rapidly growing demand for high-quality data in Large Language Models (LLMs) has intensified the need for scalable, reliable, and semantically rich data preparation pipelines. However, current practices remain dominated by ad-hoc scripts and loosely specified workflows, which lack principled abstractions, hinder reproducibility, and offer limited support for model-in-the-loop data generation. To address these challenges, we present DataFlow, a unified and extensible LLM-driven data preparation framework. DataFlow is designed with system-level abstractions that enable modular, reusable, and composable data transformations, and provides a PyTorch-style pipeline construction API for building debuggable and optimizable dataflows. The framework consists of nearly 200 reusable operators and six domain-general pipelines spanning text, mathematical reasoning, code, Text-to-SQL, agentic RAG, and large-scale knowledge extraction. To further improve usability, we introduce DataFlow-Agent, which automatically translates natural-language specifications into executable pipelines via operator synthesis, pipeline planning, and iterative verification. Across six representative use cases, DataFlow consistently improves downstream LLM performance. Our math, code, and text pipelines outperform curated human datasets and specialized synthetic baselines, achieving up to +3\% execution accuracy in Text-to-SQL over SynSQL, +7\% average improvements on code benchmarks, and 1--3 point gains on MATH, GSM8K, and AIME. Moreover, a unified 10K-sample dataset produced by DataFlow enables base models to surpass counterparts trained on 1M Infinity-Instruct data. These results demonstrate that DataFlow provides a practical and high-performance substrate for reliable, reproducible, and scalable LLM data preparation, and establishes a system-level foundation for future data-centric AI development.
Why we are recommending this paper?
Due to your Interest in: Data Science Management
Given the user's interest in data science organizations and data preparation, this paper offers a framework for scalable data pipelines, a critical component of effective data-driven strategies. The focus on LLMs for data preparation is particularly relevant to the user’s broader interests in data science.
Due to your Interest in: Data Science Management
Given the user's interest in data science organizations and data preparation, this paper offers a framework for scalable data pipelines, a critical component of effective data-driven strategies. The focus on LLMs for data preparation is particularly relevant to the user’s broader interests in data science.
ETH Zurich
AI Insights
- Combinatorial auctions: an auction format where bidders submit bids on combinations of items rather than individual items. [3]
- The paper presents a novel approach to bidding in day-ahead electricity markets using package bids and combinatorial auctions. [2]
Abstract
Bidding flexibility in day-ahead and intraday auctions would enable decentralized flexible resources, such as electric vehicles and heat pumps, to efficiently align their consumption with the intermittent generation of renewable energy. However, because these resources are individually too small to participate in those auctions directly, an aggregator (e.g., a utility) must act on their behalf. This requires aggregating many decentralized resources, which is a computationally challenging task. In this paper, we propose a computationally efficient and highly accurate method that is readily applicable to European day-ahead and intraday auctions. Distinct from existing methods, we aggregate only economically relevant power profiles, identified through price forecasts. The resulting flexibility is then conveyed to the market operator via exclusive groups of block bids. We evaluate our method for a utility serving the Swiss town of Losone, where flexibility from multiple heat pumps distributed across the grid must be aggregated and bid in the Swiss day-ahead auction. Results show that our method aggregates accurately, achieving 98% of the theoretically possible cost savings. This aggregation accuracy remains stable even as the number of heat pumps increases, while computation time grows only linearly, demonstrating strong scalability.
Bidding flexibility in day-ahead and intraday auctions would enable decentralized flexible resources, such as electric vehicles and heat pumps, to efficiently align their consumption with the intermittent generation of renewable energy. However, because these resources are individually too small to participate in those auctions directly, an aggregator (e.g., a utility) must act on their behalf. This requires aggregating many decentralized resources, which is a computationally challenging task. In this paper, we propose a computationally efficient and highly accurate method that is readily applicable to European day-ahead and intraday auctions. Distinct from existing methods, we aggregate only economically relevant power profiles, identified through price forecasts. The resulting flexibility is then conveyed to the market operator via exclusive groups of block bids. We evaluate our method for a utility serving the Swiss town of Losone, where flexibility from multiple heat pumps distributed across the grid must be aggregated and bid in the Swiss day-ahead auction. Results show that our method aggregates accurately, achieving 98% of the theoretically possible cost savings. This aggregation accuracy remains stable even as the number of heat pumps increases, while computation time grows only linearly, demonstrating strong scalability.
Why we are recommending this paper?
Due to your Interest in: Bidding
This paper’s exploration of bidding flexibility in energy auctions aligns with the user’s interest in marketing channels and bidding strategies, particularly in the context of optimizing resource allocation. The focus on decentralized flexible resources is a relevant area for strategic bidding.
Due to your Interest in: Bidding
This paper’s exploration of bidding flexibility in energy auctions aligns with the user’s interest in marketing channels and bidding strategies, particularly in the context of optimizing resource allocation. The focus on decentralized flexible resources is a relevant area for strategic bidding.
University of Bonn
Abstract
We consider the problem of searching for rays (or lines) in the half-plane. The given problem turns out to be a very natural extension of the cow-path problem that is lifted into the half-plane and the problem can also directly be motivated by a 1.5-dimensional terrain search problem. We present and analyse an efficient strategy for our setting and guarantee a competitive ratio of less than 9.12725 in the worst case and also prove a lower bound of at least 9.06357 for any strategy. Thus the given strategy is almost optimal, the gap is less than 0.06368. By appropriate adjustments for the terrain search problem we can improve on former results and present geometrically motivated proof arguments. As expected, the terrain itself can only be helpful for the searcher that competes against the unknown shortest path. We somehow extract the core of the problem.
We consider the problem of searching for rays (or lines) in the half-plane. The given problem turns out to be a very natural extension of the cow-path problem that is lifted into the half-plane and the problem can also directly be motivated by a 1.5-dimensional terrain search problem. We present and analyse an efficient strategy for our setting and guarantee a competitive ratio of less than 9.12725 in the worst case and also prove a lower bound of at least 9.06357 for any strategy. Thus the given strategy is almost optimal, the gap is less than 0.06368. By appropriate adjustments for the terrain search problem we can improve on former results and present geometrically motivated proof arguments. As expected, the terrain itself can only be helpful for the searcher that competes against the unknown shortest path. We somehow extract the core of the problem.
AI Insights
- They show that this adjusted strategy attains the same X-coordinates in the free space as the original part of the path, while potentially having a larger Y-coordinate. [3]
- The authors use a budget-based approach to ensure that the adjusted strategy has the same path length as the original part, and they calculate the return points during the movement to guarantee this property. [3]
- They also show that the adjusted strategy is optimal in terms of path length. [3]
- The authors use mathematical techniques such as calculus and optimization to derive the results. [3]
- The paper discusses the application and adaptation of the cow path strategy to the terrain search problem, highlighting the importance of exploiting the terrain's slope to minimize the path length. [3]
- cow path problem: a classic problem in computational geometry where an agent must find the shortest path between two points while avoiding obstacles. [3]
- The paper presents a new strategy for searching in a terrain with obstacles, which is an extension of the traditional cow path problem. [2]
- pure strategy: a starting point for the adjusted strategy, which is used as a basis for the adjustments made according to the terrain. [1]
Why we are recommending this paper?
Due to your Interest in: Paid Search
Due to your Interest in: Paid Search
arXiv
Abstract
The pursuit of high-performance data transfer often focuses on raw network bandwidth, and international links of 100 Gbps or higher are frequently considered the primary enabler. While necessary, this network-centric view is incomplete, equating provisioned link speeds with practical, sustainable data movement capabilities across the entire edge-to-core spectrum. This paper investigates six common paradigms, from the often-cited constraints of network latency and TCP congestion control algorithms to host-side factors such as CPU performance and virtualization that critically impact data movement workflows. We validated our findings using a latency-emulation-capable testbed for high-speed WAN performance prediction and through extensive production measurements from resource-constrained edge environments to a 100 Gbps operational link connecting Switzerland and California, U.S. These results show that the principal bottlenecks often reside outside the network core, and that a holistic hardware-software co-design ensures consistent performance, whether moving data at 1 Gbps or 100 Gbps and faster. This approach effectively closes the fidelity gap between benchmark results and diverse and complex production environments.
The pursuit of high-performance data transfer often focuses on raw network bandwidth, and international links of 100 Gbps or higher are frequently considered the primary enabler. While necessary, this network-centric view is incomplete, equating provisioned link speeds with practical, sustainable data movement capabilities across the entire edge-to-core spectrum. This paper investigates six common paradigms, from the often-cited constraints of network latency and TCP congestion control algorithms to host-side factors such as CPU performance and virtualization that critically impact data movement workflows. We validated our findings using a latency-emulation-capable testbed for high-speed WAN performance prediction and through extensive production measurements from resource-constrained edge environments to a 100 Gbps operational link connecting Switzerland and California, U.S. These results show that the principal bottlenecks often reside outside the network core, and that a holistic hardware-software co-design ensures consistent performance, whether moving data at 1 Gbps or 100 Gbps and faster. This approach effectively closes the fidelity gap between benchmark results and diverse and complex production environments.
AI Insights
- The common paradigm that powerful CPUs are essential for high transfer rates is not supported by evidence. [3]
- Software efficiency and storage architecture matter more than CPU raw computing power. [3]
- High-speed data transfer Data movement appliances System control accessible to the user The right software makes adequate CPUs perform exceptionally; the wrong software cannot be saved by the most powerful CPUs. [3]
- The common paradigm that powerful CPUs are essential for high transfer rates is not supported by evidence. [3]
- Virtualization and cloud abstraction impose significant performance penalties on data movement. [2]
Why we are recommending this paper?
Due to your Interest in: Direction on Data Science Organizations
Due to your Interest in: Direction on Data Science Organizations
Mantis Analytics
Abstract
Large language models can already query databases, yet most existing systems remain reactive: they rely on explicit user prompts and do not actively explore data. We introduce DAR (Data Agnostic Researcher), a multi-agent system that performs end-to-end database research without human-initiated queries. DAR orchestrates specialized AI agents across three layers: initialization (intent inference and metadata extraction), execution (SQL and AI-based query synthesis with iterative validation), and synthesis (report generation with built-in quality control). All reasoning is executed directly inside BigQuery using native generative AI functions, eliminating data movement and preserving data governance. On a realistic asset-incident dataset, DAR completes the full analytical task in 16 minutes, compared to 8.5 hours for a professional analyst (approximately 32x times faster), while producing useful pattern-based insights and evidence-grounded recommendations. Although human experts continue to offer deeper contextual interpretation, DAR excels at rapid exploratory analysis. Overall, this work shifts database interaction from query-driven assistance toward autonomous, research-driven exploration within cloud data warehouses.
Large language models can already query databases, yet most existing systems remain reactive: they rely on explicit user prompts and do not actively explore data. We introduce DAR (Data Agnostic Researcher), a multi-agent system that performs end-to-end database research without human-initiated queries. DAR orchestrates specialized AI agents across three layers: initialization (intent inference and metadata extraction), execution (SQL and AI-based query synthesis with iterative validation), and synthesis (report generation with built-in quality control). All reasoning is executed directly inside BigQuery using native generative AI functions, eliminating data movement and preserving data governance. On a realistic asset-incident dataset, DAR completes the full analytical task in 16 minutes, compared to 8.5 hours for a professional analyst (approximately 32x times faster), while producing useful pattern-based insights and evidence-grounded recommendations. Although human experts continue to offer deeper contextual interpretation, DAR excels at rapid exploratory analysis. Overall, this work shifts database interaction from query-driven assistance toward autonomous, research-driven exploration within cloud data warehouses.
Why we are recommending this paper?
Due to your Interest in: Data Science Management
Due to your Interest in: Data Science Management
UC San Diego
Abstract
Prosody -- the melody of speech -- conveys critical information often not captured by the words or text of a message. In this paper, we propose an information-theoretic approach to quantify how much information is expressed by prosody alone and not by text, and crucially, what that information is about. Our approach applies large speech and language models to estimate the mutual information between a particular dimension of an utterance's meaning (e.g., its emotion) and any of its communication channels (e.g., audio or text). We then use this approach to quantify how much information is conveyed by audio and text about sarcasm, emotion, and questionhood, using speech from television and podcasts. We find that for sarcasm and emotion the audio channel -- and by implication the prosodic channel -- transmits over an order of magnitude more information about these features than the text channel alone, at least when long-term context beyond the current sentence is unavailable. For questionhood, prosody provides comparatively less additional information. We conclude by outlining a program applying our approach to more dimensions of meaning, communication channels, and languages.
Prosody -- the melody of speech -- conveys critical information often not captured by the words or text of a message. In this paper, we propose an information-theoretic approach to quantify how much information is expressed by prosody alone and not by text, and crucially, what that information is about. Our approach applies large speech and language models to estimate the mutual information between a particular dimension of an utterance's meaning (e.g., its emotion) and any of its communication channels (e.g., audio or text). We then use this approach to quantify how much information is conveyed by audio and text about sarcasm, emotion, and questionhood, using speech from television and podcasts. We find that for sarcasm and emotion the audio channel -- and by implication the prosodic channel -- transmits over an order of magnitude more information about these features than the text channel alone, at least when long-term context beyond the current sentence is unavailable. For questionhood, prosody provides comparatively less additional information. We conclude by outlining a program applying our approach to more dimensions of meaning, communication channels, and languages.
AI Insights
- The study of prosody and its relationship with emotions and communication has a rich history, dating back decades. [3]
- Information-theoretic models have been proposed as a way to quantify the complexity of language processing and memory effects in sentence comprehension. [3]
- Multimodality: A social semiotic approach to communication that considers multiple modes of expression, including speech, gesture, and facial expressions. [3]
- Information-theoretic models: Mathematical frameworks for quantifying the complexity of language processing and memory effects in sentence comprehension. [3]
- The study of prosody has contributed significantly to our understanding of human communication, particularly in relation to emotions and intentions. [3]
- Information-theoretic models offer a new perspective on language processing, highlighting the importance of memory effects and complexity in sentence comprehension. [3]
- Limited scope: The study focuses primarily on prosody and its relationship with emotions and communication, neglecting other aspects of human interaction. [3]
- Recent research has focused on the use of multimodal approaches, incorporating both speech and non-speech cues, to better understand human communication. [2]
Why we are recommending this paper?
Due to your Interest in: Marketing Channels
Due to your Interest in: Marketing Channels
Meituan
Abstract
Recently, joint advertising has gained significant attention as an effective approach to enhancing the efficiency and revenue of advertising slot allocation. Unlike traditional advertising, which allocates advertising slots exclusively to a single advertiser, joint advertising displays advertisements from brands and stores that have established a joint selling relationship within the same advertising slot. However, existing approaches often struggle to accommodate both joint and traditional advertising frameworks, thereby limiting the revenue potential and generalizability of joint advertising. Furthermore, these methods are constrained by two critical limitations: they generally neglect the influence of global externalities, and they fail to address the bidding variability stemming from multi-party advertiser participation. Collectively, these limitations present substantial challenges to the design of joint auction mechanisms. To address these challenges, we propose a Joint Auction Framework incorporating Externalities and Adaptation, and leverage the automated mechanism design (AMD) method through our proposed JEANet to compute joint auction mechanisms that satisfy the conditions of individual rationality (IR) and approximate dominant strategy incentive compatibility (DSIC). As the first AMD method to integrate global externalities into joint auctions, JEANet dynamically adapts to the bidding characteristics of multi-party advertiser and enables unified auctions that integrate both joint and traditional advertising. Extensive experimental results demonstrate that JEANet outperforms state-of-the-art baselines in multi-slot joint auctions.
Recently, joint advertising has gained significant attention as an effective approach to enhancing the efficiency and revenue of advertising slot allocation. Unlike traditional advertising, which allocates advertising slots exclusively to a single advertiser, joint advertising displays advertisements from brands and stores that have established a joint selling relationship within the same advertising slot. However, existing approaches often struggle to accommodate both joint and traditional advertising frameworks, thereby limiting the revenue potential and generalizability of joint advertising. Furthermore, these methods are constrained by two critical limitations: they generally neglect the influence of global externalities, and they fail to address the bidding variability stemming from multi-party advertiser participation. Collectively, these limitations present substantial challenges to the design of joint auction mechanisms. To address these challenges, we propose a Joint Auction Framework incorporating Externalities and Adaptation, and leverage the automated mechanism design (AMD) method through our proposed JEANet to compute joint auction mechanisms that satisfy the conditions of individual rationality (IR) and approximate dominant strategy incentive compatibility (DSIC). As the first AMD method to integrate global externalities into joint auctions, JEANet dynamically adapts to the bidding characteristics of multi-party advertiser and enables unified auctions that integrate both joint and traditional advertising. Extensive experimental results demonstrate that JEANet outperforms state-of-the-art baselines in multi-slot joint auctions.
Why we are recommending this paper?
Due to your Interest in: Bidding
Due to your Interest in: Bidding
University of Florida
Abstract
Large language models (LLMs) exhibit remarkable capabilities, yet their reasoning remains opaque, raising safety and trust concerns. Attribution methods, which assign credit to input features, have proven effective for explaining the decision making of computer vision models. From these, context attributions have emerged as a promising approach for explaining the behavior of autoregressive LLMs. However, current context attributions produce incomplete explanations by directly relating generated tokens to the prompt, discarding inter-generational influence in the process. To overcome these shortcomings, we introduce the Context Attribution via Graph Explanations (CAGE) framework. CAGE introduces an attribution graph: a directed graph that quantifies how each generation is influenced by both the prompt and all prior generations. The graph is constructed to preserve two properties-causality and row stochasticity. The attribution graph allows context attributions to be computed by marginalizing intermediate contributions along paths in the graph. Across multiple models, datasets, metrics, and methods, CAGE improves context attribution faithfulness, achieving average gains of up to 40%.
Large language models (LLMs) exhibit remarkable capabilities, yet their reasoning remains opaque, raising safety and trust concerns. Attribution methods, which assign credit to input features, have proven effective for explaining the decision making of computer vision models. From these, context attributions have emerged as a promising approach for explaining the behavior of autoregressive LLMs. However, current context attributions produce incomplete explanations by directly relating generated tokens to the prompt, discarding inter-generational influence in the process. To overcome these shortcomings, we introduce the Context Attribution via Graph Explanations (CAGE) framework. CAGE introduces an attribution graph: a directed graph that quantifies how each generation is influenced by both the prompt and all prior generations. The graph is constructed to preserve two properties-causality and row stochasticity. The attribution graph allows context attributions to be computed by marginalizing intermediate contributions along paths in the graph. Across multiple models, datasets, metrics, and methods, CAGE improves context attribution faithfulness, achieving average gains of up to 40%.
AI Insights
- Contextualized attention weights: These represent the importance of each input token in generating a particular output. [3]
- Attribution methods: These aim to identify which parts of the input data contribute most to a model's predictions or generation. [3]
- It provides a more accurate representation of how LLMs generate text, which can be useful for tasks such as debugging, fine-tuning, or even generating new content. [3]
- The proposed method may not generalize well to other types of models or tasks. [3]
- It requires significant computational resources and memory to compute the contextualized attention weights. [3]
- The paper proposes a new method for explaining the behavior of large language models (LLMs) using contextualized attention weights. [2]
- The authors argue that previous methods for attributing model generation to context are limited in their ability to capture the complex interactions between input tokens and the model's internal state. [1]
Why we are recommending this paper?
Due to your Interest in: Attribution
Due to your Interest in: Attribution
Johns Hopkins University
Abstract
Forensic scientists often need to identify an unknown speaker or writer in cases such as ransom calls, covert recordings, alleged suicide notes, or anonymous online communications, among many others. Speaker recognition in the speech domain usually examines phonetic or acoustic properties of a voice, and these methods can be accurate and robust under certain conditions. However, if a speaker disguises their voice or employs text-to-speech software, vocal properties may no longer be reliable, leaving only their linguistic content available for analysis. Authorship attribution methods traditionally use syntactic, semantic, and related linguistic information to identify writers of written text (authorship attribution). In this paper, we apply a content-based authorship approach to speech that has been transcribed into text, using what a speaker says to attribute speech to individuals (speaker attribution). We introduce a stylometric method, StyloSpeaker, which incorporates character, word, token, sentence, and style features from the stylometric literature on authorship, to assess whether two transcripts were produced by the same speaker. We evaluate this method on two types of transcript formatting: one approximating prescriptive written text with capitalization and punctuation and another normalized style that removes these conventions. The transcripts' conversation topics are also controlled to varying degrees. We find generally higher attribution performance on normalized transcripts, except under the strongest topic control condition, in which overall performance is highest. Finally, we compare this more explainable stylometric model to black-box neural approaches on the same data and investigate which stylistic features most effectively distinguish speakers.
Forensic scientists often need to identify an unknown speaker or writer in cases such as ransom calls, covert recordings, alleged suicide notes, or anonymous online communications, among many others. Speaker recognition in the speech domain usually examines phonetic or acoustic properties of a voice, and these methods can be accurate and robust under certain conditions. However, if a speaker disguises their voice or employs text-to-speech software, vocal properties may no longer be reliable, leaving only their linguistic content available for analysis. Authorship attribution methods traditionally use syntactic, semantic, and related linguistic information to identify writers of written text (authorship attribution). In this paper, we apply a content-based authorship approach to speech that has been transcribed into text, using what a speaker says to attribute speech to individuals (speaker attribution). We introduce a stylometric method, StyloSpeaker, which incorporates character, word, token, sentence, and style features from the stylometric literature on authorship, to assess whether two transcripts were produced by the same speaker. We evaluate this method on two types of transcript formatting: one approximating prescriptive written text with capitalization and punctuation and another normalized style that removes these conventions. The transcripts' conversation topics are also controlled to varying degrees. We find generally higher attribution performance on normalized transcripts, except under the strongest topic control condition, in which overall performance is highest. Finally, we compare this more explainable stylometric model to black-box neural approaches on the same data and investigate which stylistic features most effectively distinguish speakers.
Why we are recommending this paper?
Due to your Interest in: Attribution
Due to your Interest in: Attribution
Interests not found
We did not find any papers that match the below interests.
Try other terms also consider if the content exists in arxiv.org.
- customer relationship management (crm) optimization
You can edit or add more interests any time.
Help us improve your experience!
This project is on its early stages your feedback can be pivotal on the future of the project.
Let us know what you think about this week's papers and suggestions!
Give Feedback