论文检索
输入标题、作者或关键词，从 100,903 篇学术成果中精准定位
Online advertising platforms serve as a critical bridge between advertisers and media, requiring precise prediction of user behaviors. Systematic underestimation or overestimation in these predictions can undermine the interests of both parties. Existing calibration methods fall short in addressing two key challenges. First, significant differences in CVR distributions across various targets lead to biased calibration when using global posterior statistics, causing some sample groups to be overestimated while others are underestimated. Second, current field-aware approaches are typically limited to single-field calibration and fail to account for field sensitivity. To overcome these limitations, we propose a Pareto Frontier-based Multi-field Personalized Calibration (PF-MPC) method which formulates multi-field calibration as a multi-objective optimization problem. PF-MPC identifies the optimal Pareto-efficient weight combinations to balance the conflicting calibration errors across different fields. We evaluate PF-MPC on both public calibration benchmark and a large-scale industrial dataset. Experimental results demonstrate that our method achieves significant improvements in calibration performance compared to existing approaches.
In this work, we present a new random sampling method for data streams where the probability of an element's inclusion in the sample is proportional to a weight associated with that element. Our method is based on sampling with replacement, although most of the literature on this topic has focused on sampling without replacement. Our algorithm generates a weighted random sample in one pass over a population of unknown size. At any point in time, the sample is representative of the population seen so far and can be directly used by other modules without requiring any post-processing. We formally prove the correctness and efficiency of our method. An experimental analysis shows the performance of our method in practice when compared to state-of-the-art methods.
Graph anomaly detection (GAD), which aims to identify abnormal nodes that deviate from the majority, has become increasingly important in high-stakes Web domains. However, existing GAD methods follow a ''one model per dataset'' paradigm, leading to high computational costs, substantial data demands, and poor generalization when transferred to new datasets. This calls for a foundation model that enables a ''one-for-all'' GAD solution capable of detecting anomalies across diverse graphs without retraining. Yet, achieving this is challenging due to the large structural and feature heterogeneity across domains. In this paper, we propose TFM4GAD, a simple yet effective framework that adapts tabular foundation models (TFMs) for graph anomaly detection. Our key insight is that the core challenges of foundation GAD, handling heterogeneous features, generalizing across domains, and operating with scarce labels, are the exact problems that modern TFMs are designed to solve via synthetic pre-training and powerful in-context learning. The primary challenge thus becomes structural: TFMs are agnostic to graph topology. TFM4GAD bridges this gap by ''flattening'' the graph, constructing an augmented feature table that enriches raw node features with Laplacian embeddings, local and global structural characteristics, and anomaly-sensitive neighborhood aggregations. This augmented table is processed by a TFM in a fully in-context regime. Extensive experiments on multiple datasets with various TFM backbones reveal that TFM4GAD surprisingly achieves significant performance gains over specialized GAD models trained from scratch. Our work offers a new perspective and a practical paradigm for leveraging TFMs as powerful, generalist graph anomaly detectors.
User behavior on social media---from scrolling and viewing to liking, reposting, and posting---yet most research relies on self-reports that obscure fine-grained usage patterns. We analyze high-resolution activity logs from 209 U.S. X (Twitter) users tracked over four weeks to identify distinct behavioral profiles based on session-level features. Latent profile analysis reveals three groups---interactors (32.52%), lurkers (60.45%), and creators (7.03%) that differ in engagement intensity, demographics, and content exposure. Interactors and lurkers skew younger and Democratic, whereas creators skew older and more Republican, consuming more ideological and low-credibility content. These results link behavioral heterogeneity to systematically different information environments and suggest that platform interventions may operate unevenly across user types. Our findings demonstrate the value of log-based behavioral segmentation for understanding online participation and motivate profile-aware platform governance and content moderation strategies.
Recent advances in clinical prediction leverage large language models (LLMs) to extract semantic information from Electronic Health Records (EHRs). However, LLMs could produce biased or hallucinated responses camouflaged by their fluency and realistic appearance, which is unacceptable in diagnosis prediction. Uncertainty estimation (UE) has emerged as an effective approach to address this challenge by quantifying hallucination levels and prediction confidence in LLM outputs. Yet, directly determining diagnosis predictions based on UE remains insufficient, as diagnoses with high uncertainty may still correspond to correct outcomes. To this end, we propose ULoR, an uncertainty-aware leave-one-out refinement framework for reliable diagnosis prediction. Specifically, we first compute the UE scores by integrating statistical information from multiple samples of the model's diagnosis ranking distributions and leverage these scores for initial predictions. Then, guided by the leave-one-out strategy, we construct multiple-choice tasks for high-uncertainty diagnoses using external syndrome knowledge and fine-tune the refinement component to resolve them, thereby confirming or replacing uncertain predictions. Extensive experiments on two real-world EHR datasets demonstrate that ULoR consistently outperforms state-of-the-art baselines, showcasing its practical utility in real-world clinical settings.
Table retrieval is the task of retrieving the most relevant tables from large-scale corpora given natural language queries. However, structural and semantic discrepancies between unstructured text and structured tables make embedding alignment particularly challenging. Recent methods such as QGpT attempt to enrich table semantics by generating synthetic queries, yet they still rely on coarse partial-table sampling and simple fusion strategies, which limit semantic diversity and hinder effective query–table alignment. We propose STAR (Semantic Table Representation), a lightweight framework that improves semantic table representation through semantic clustering and weighted fusion. STAR first applies header-aware K-means clustering to group semantically similar rows and selects representative centroid instances to construct a diverse partial table. It then generates cluster-specific synthetic queries to comprehensively cover the table's semantic space. Finally, STAR employs weighted fusion strategies to integrate table and query embeddings, enabling fine-grained semantic alignment. This design enables STAR to capture complementary information from structured and textual sources, improving the expressiveness of table representations. Experiments on five benchmarks show that STAR achieves consistently higher Recall than QGpT on all datasets, demonstrating the effectiveness of semantic clustering and weighted fusion for robust table representation. Our code is available at https://github.com/adsl135789/STAR.
Foundation models for tabular data, such as the Tabular Prior-data Fitted Network (TabPFN), are pre-trained on a massive number of synthetic datasets generated by structural causal models (SCM). They leverage in-context learning to offer high predictive accuracy in real-world tasks. However, the fairness properties of these foundational models, which incorporate ideas from causal reasoning during pre-training, remain underexplored. In this work, we conduct a comprehensive empirical evaluation of TabPFN and its fine-tuned variants, assessing predictive performance, fairness, and robustness across varying dataset sizes and distributional shifts. Our results reveal that while TabPFN achieves stronger predictive accuracy compared to baselines and exhibits robustness to spurious correlations, improvements in fairness are moderate and inconsistent, particularly under missing-not-at-random (MNAR) covariate shifts. These findings suggest that the causal pre-training in TabPFN is helpful but insufficient for algorithmic fairness, highlighting implications for deploying TabPFN (and similar) models in practice and the need for further fairness interventions.
Academic question answering requires reasoning over heterogeneous scholarly graphs, where queries range from simple attribute lookups to multi-hop inference across author--paper--venue structures. Existing retrieval-augmented generation (RAG) systems struggle in this setting due to three limitations: (1) fixed retrieval strategies that do not adapt to varying query complexity, (2) the absence of sufficiency evaluation leading to incomplete or misaligned evidence, and (3) a lack of structured verification against graph facts. To address these issues, we propose an agentic heterogeneous graph RAG method that transforms the three core stages of the RAG pipeline into explicit agentic decision steps. A query-aware retrieval agent analyzes query type and selects an appropriate graph traversal strategy; a sufficiency-aware reranking agent assesses evidence completeness and adaptively expands the retrieved subgraph; and a graph-grounded verification agent checks entity, relation, and attribute correctness before finalizing the answer. Experiments on heterogeneous graphs constructed from OpenAlex and DBLP suggest that our method consistently outperforms strong LLM, graph-augmented RAG, and agent-based baselines.
We present ReSuMe, a general framework for mutual enhancement of dense retrieval systems and document summarizers through reinforcement learning. The framework jointly optimizes a language model for generating retrieval-oriented summaries and adapts the retrieval model to these summaries through alternating fine-tuning phases. We employ Group Relative Policy Optimization (GRPO) to fine-tune the language model based on retrieval relevance rather than linguistic quality alone, while the retrieval model is iteratively updated using contrastive learning on the generated summaries. This co-optimization process addresses the fundamental distribution shift problem that arises when retrieval models trained on full documents must operate on synthetic summaries during inference. By progressively reducing this distribution gap, our framework yields two key benefits: improved retrieval performance and a high-quality document summarizer optimized for retrieval tasks. We demonstrate our framework using Contriever on the MS-MARCO dataset, achieving consistent improvements of 13.2% in MRR@10 and 6.7% in Recall@100 over the baseline. The framework is model-agnostic and can be applied to enhance any dense retrieval system while simultaneously producing an effective document summarization model.
Large Language Models (LLMs) perform well across many tasks but degrade when processing large collections of repetitive or highly similar inputs, a common scenario in applications such as near-duplicate search results and large e-commerce catalogs. In these settings, concatenation-based approaches—long-context prompting and supervised fine-tuning—suffer from attention saturation and diminished signal-to-noise ratio, causing models to miss subtle but important distinctions as input size grows. We introduce SEER (Set Encoding for Efficient Representation), a framework that enables LLMs to handle massive sets of near-duplicate items through a single learned token. SEER first encodes individual items with a pretrained embedding model, then aggregates them using an adapter that captures inter-item relationships and preserves fine-grained differences while mitigating redundancy. To ensure both discriminative and generative capabilities, we propose a multi-task alignment strategy that supervises set-level descriptions across multiple semantic dimensions. Experiments on a large-scale e-commerce dataset demonstrate that SEER substantially outperforms in-context and fine-tuned LLM baselines, maintaining stable performance even when processing thousands of highly similar items. These results establish SEER as an effective and scalable approach for LLM processing of dense, redundant input sets.
Web searchers continuously form expectations about document content based on snippets and titles, yet when these expectations are violated, their attention and satisfaction are disrupted. Detecting such expectancy violations in real-time can enable adaptive, user-aware Web systems that respond to cognitive mismatches. This paper investigates whether eye-tracking features can predict expectancy violations during Web search and identifies which temporal aspects of attention carry predictive information. Using data from 34 participants performing controlled search tasks, we extracted four gaze metrics, i.e. time to first fixation, total fixation duration, number of fixations, and mean fixation duration, and trained machine learning models using a leave-one-participant-out cross-validation approach. Sustained attention features, particularly total fixation duration and number of fixations, predicted expectancy violations with 62.6% accuracy (p = .008), while initial attention metrics performed at chance. The results reveal that expectancy violations manifest through extended visual engagement rather than immediate orienting responses. The findings of this work contribute to the theoretical understanding of user–system interaction on the Web and provide a foundation for adaptive retrieval interfaces capable of detecting cognitive surprise and delivering timely support.
Topic modeling plays a critical role in organizing and understanding large-scale web content. While neural topic models (NTMs) based on variational autoencoders (VAEs) have achieved notable success in analyzing textual data, they remain limited in addressing the multimodal nature of modern web content. Existing unimodal or multimodal extensions often suffer from posterior collapse and fail to capture the directional semantics inherent in both text and images, resulting in incoherent topics and limited interpretability. To address these challenges, we propose MM-vNTM (MultiModal Neural Topic Model with von Mises-Fisher Mixtures), a framework for web-scale topic discovery over multimodal data. MM-vNTM leverages pre-aligned cross-modal embeddings as inputs and jointly models document-level representations of text and image modalities in a shared hyperspherical latent space. Furthermore, it defines topics as mixtures of von Mises-Fisher (vMF) distributions in the L2-normalized word embedding space, explicitly capturing directional similarity. Experiments on multimedia web datasets demonstrate that MM-vNTM consistently outperforms state-of-the-art unimodal and multimodal baselines in terms of overall topic quality, highlighting its effectiveness for real-world web scenarios.
The rapid proliferation of AI-generated content on the Web presents a structural risk to information retrieval, as search engines and Retrieval-Augmented Generation (RAG) systems increasingly consume evidence produced by the Large Language Models (LLMs). We characterize this ecosystem-level failure mode as Retrieval Collapse, a two-stage process where (1) AI-generated content dominates search results, eroding source diversity, and (2) low-quality or adversarial content infiltrates the retrieval pipeline. We analyzed this dynamic through controlled experiments involving both high-quality SEO-style content and adversarially crafted content. In the SEO scenario, a 67% pool contamination led to over 80% exposure contamination, creating a homogenized yet deceptively healthy state where answer accuracy remains stable despite the reliance on synthetic sources. Conversely, under adversarial contamination, baselines like BM25 exposed ~19% of harmful content, whereas LLM-based rankers demonstrated stronger suppression capabilities. These findings highlight the risk of retrieval pipelines quietly shifting toward synthetic evidence and the need for retrieval-aware strategies to prevent a self-reinforcing cycle of quality decline in Web-grounded systems.
Most graph unlearning evaluations delete a random fraction k% of nodes or edges. In practice, however, controllers act on natural?language erasure requests. To support evaluation under such text inputs, we release Text2Forget, a DSAR-style natural-language erasure-request corpus and benchmark for CoraFull, PubMed, and Yelp, together with a resolver that maps request text to concrete graph targets. Our pipeline designs law?motivated, retrieval-grounded requests that use human-readable anchors and resolves each request to real nodes and edges via a hybrid BM25-plus-embedding index, yielding thousands of requests per dataset. Because this corpus is large, a deterministic ranking score selects a small, actionable top-k subset for evaluation. A plug-and-play harness (GCN, GAT, and GIN with UtU, GNNDelete, GIF, and Retrain) reports privacy via membership-inference AUC (MI-AUC) on forgotten items, averaged over the selected requests and backbones. Across datasets, request-driven forget-sets induce membership-inference and task-accuracy behavior broadly comparable to standard random-forget and retraining baselines, indicating that the Text2Forget corpus can serve as a practical benchmark within existing privacy–utility trade-offs. The released package includes the corpus, resolved targets, confidence scores, seeds, and scripts, supporting reproducible, text-driven unlearning studies. We release our package at https://github.com/ImranAhsan23/Text2Forget.
The performance of large language model (LLM) inference engines heavily relies on deployment-specific parameters such as batch size, parallelism, CUDA graphs, and GPU and network topology, however, with every new LLM released, the optimal setting of these parameters becomes a problem of interest. Additionally, as developers find new applications for LLMs, the parameter space only increases, with large scale inference deployments that span multiple nodes (i.e. llm-d [5]). In this work, we demonstrate ''llm-tuna'', an open-source framework that automates parameter search for vLLM [4] based inference workloads across a diverse set of hardware setups. Our framework utilizes Bayesian optimization via Optuna to select the best configuration while parallelizing trials across multiple nodes to enable large scale studies. We evaluate our system on a mixture-of-experts (MoE) model family (Qwen3-30B-A3B, Kimi-K2-INT4, GPT-OSS-120B, Deepseek-R1-671B), configuring each experiment to avoid artifacts of KV-cache pre-emption effects. Our method achieves output throughput gains of up to 32.9% while never suggesting configurations that are worse than vLLMs' baseline performance. Our results demonstrate the potential of systematic auto-tuning to expose non-obvious performance regimes and guide operator defaults.
Despite significant progress in text anomaly detection for web applications such as spam filtering and fake news detection, existing methods are fundamentally limited to document-level analysis, unable to identify which specific parts of a text are anomalous. We introduce token-level anomaly detection, a novel paradigm that enables fine-grained localization of anomalies within text. We formally define text anomalies at both document and token-levels, and propose a unified detection framework that operates across multiple levels. To facilitate research in this direction, we collect and annotate three benchmark datasets spanning spam, reviews and grammar errors with token-level labels. Experimental results demonstrate that our framework achieves better performance than other 6 baselines, opening new possibilities for precise anomaly localization in text. All the codes and data are publicly available on https://github.com/charles-cao/TokenCore.
Scientific fact-checking is vital for assessing claims in specialized domains such as biomedicine and materials science, yet existing systems often hallucinate or apply inconsistent reasoning, especially when verifying technical, compositional claims against an evidence snippet under source and cost/latency constraints. We present a pipeline centered on atomic predicate–argument decomposition and calibrated, uncertainty-gated corroboration: atomic facts are aligned to local snippets via embeddings, verified by a compact evidence-grounded checker, and only facts with uncertain support trigger domain-restricted web search over authoritative sources. The system supports both binary and tri-valued classification where it predicts labels from Supported, Refuted, NEI for three-way tasks. We evaluate under two regimes, Context-Only (no web) and Context+Web (uncertainty-gated web corroboration); when retrieved evidence conflicts with the provided context, we abstain with NEI rather than overriding the context. On multiple benchmarks, our framework surpasses the strongest benchmarks. In our experiments, web corroboration was invoked for only a minority of atomic facts on average, indicating that external evidence is consulted selectively under calibrated uncertainty rather than routinely. Overall, coupling atomic granularity with calibrated, uncertainty-gated corroboration yields more interpretable and context-conditioned verification, making the approach well-suited to high-stakes, single-document settings that demand traceable rationales, predictable cost/latency, and conservative abstention under domain and evidence grounding constraints.
Trending news detection in low-traffic search environments faces a fundamental cold-start problem, where a lack of query volume prevents systems from identifying emerging or long-tail trends. Existing methods relying on keyword frequency or query spikes are inherently slow and ineffective in these sparse settings, lagging behind real-world shifts in attention. We introduce RTTP, a novel Real-Time Trending Prediction framework that generates search queries directly from news content instead of waiting for users to issue them. RTTP leverages a continual learning LLM (CL-LLM) that converts posts into search-style queries and scores them using engagement strength + creator authority, enabling early trend surfacing before search volume forms. To ensure adaptation without degrading reasoning, we propose Mix-Policy DPO, a new preference-based continual learning approach that combines on-policy stability with off-policy novelty to mitigate catastrophic forgetting during model upgrades. Deployed at production scale on Facebook and Meta AI products, RTTP delivers +91.4% improvement in tail-trend detection precision@500 and +19% query generation accuracy over industry baselines, while sustaining stable performance after multi-week online training. This work demonstrates that LLM-generated synthetic search signals, when aligned and continually updated, unlock timely trend understanding in low-traffic search environments.
The Form 10-K provides a detailed overview of a company's operations, strategies, competitive environment, and future outlook. Investors meticulously analyse annual changes in these reports to identify subtle shifts in disclosed information. However, financial narratives often contain extensive boilerplate text, which dilutes the efficacy of traditional sentiment analysis. We propose a novel edit-level analysis framework to isolate meaningful changes in Management Discussion and Analysis (MD&A) sections of 10-K filings. Using sentence embeddings, we classify each sentence in the filings for years t and t+1 as Kept (repeated boilerplate), Deleted (removed information), and Added (newly introduced information). We then construct a net sentiment change signal from the difference in average FinBERT-tone sentiment between Added and Deleted content. This Core Sentiment measure has been demonstrated to outperform conventional bag-of-words sentiment measures. Furthermore, based on cleaned and standardised 10-K corpora (e.g., EDGAR-CORPUS) and our Flex10K pre-processing pipeline, our approach provides a reproducible basis for quantifying changes between annual reports.