In recent years, deep clustering has achieved encouraging results. However, existing deep clustering methods work with the traditional Euclidean space and thus present deficiency on clustering complex structures. On the contrary, Riemannian geometry provides an elegant framework to model complex structures as well as a powerful tool for clustering, i.e., the Ricci flow. In this paper, we rethink the problem of deep clustering, and introduce the Riemannian geometry to deep clustering for the first time. Deep clustering in Riemannian manifold still faces significant challenges: (1) Ricci flow itself is unaware of cluster membership, (2) Ricci curvature prevents the gradient backpropagation, and (3) learning the flow largely remains open in the manifold. To bridge these gaps, we propose a novel Riemannian generative model (RicciNet), a neural Ricci flow with several theoretical guarantees. The novelty is that we model the dynamic self-clustering process of Ricci flow: data points move to the respective clusters in the manifold, influenced by Ricci curvatures. The point's trajectory is characterized by a parametric velocity, taking the form of Ordinary Differential Equation (ODE). Specifically, we encode data points as samples of Gaussian mixture in the manifold where we propose two types of reparameterization approaches: Gumbel reparameterization, and geometric trick. We formulate a differentiable Ricci curvature parameterized by a Riemannian graph convolution. Thereafter, we propose a geometric learning approach in which we study the geometric regularity of the point's trajectory, and learn the flow via distance matching and velocity matching. Consequently, data points go along the shortest Ricci flow to complete clustering. Extensive empirical results show RicciNet outperforms Euclidean deep methods.
论文检索
输入标题、作者或关键词，从 432 篇学术成果中精准定位
Web-based applications such as chatbots, search engines and news recommendations continue to grow in scale and complexity with the recent surge in the adoption of large language models (LLMs). Online model selection has thus garnered increasing attention due to the need to choose the best model among a diverse set while balancing task reward and exploration cost. Organizations faces decisions like whether to employ a costly API-based LLM or a locally finetuned small LLM, weighing cost against performance. Traditional selection methods often evaluate every candidate model before choosing one, which are becoming impractical given the rising costs of training and finetuning LLMs. Moreover, it is undesirable to allocate excessive resources towards exploring poor-performing models. While some recent works leverage online bandit algorithm to manage such exploration-exploitation trade-off in model selection, they tend to overlook the increasing-then-converging trend in model performances as the model is iteratively finetuned, leading to less accurate predictions and suboptimal model selections. In this paper, we propose a time-increasing bandit algorithm TI-UCB, which effectively predicts the increase of model performances due to training or finetuning and efficiently balances exploration and exploitation in model selection. To further capture the converging points of models, we develop a change detection mechanism by comparing consecutive increase predictions. We theoretically prove that our algorithm achieves a lower regret upper bound, improving from prior works' polynomial regret to logarithmic in a similar setting. The advantage of our method is also empirically validated through extensive experiments on classification model selection and online selection of LLMs. Our results highlight the importance of utilizing increasing-then-converging pattern for more efficient and economic model selection in the deployment of LLMs.
LinkNER: Linking Local Named Entity Recognition Models to Large Language Models using Uncertainty
PDF ↗Named Entity Recognition (NER) serves as a fundamental task in natural language understanding, bearing direct implications for web content analysis, search engines, and information retrieval systems. Fine-tuned NER models exhibit satisfactory performance on standard NER benchmarks. However, due to limited fine-tuning data and lack of knowledge, it performs poorly on unseen entity recognition. As a result, the usability and reliability of NER models in web-related applications are compromised. Instead, Large Language Models (LLMs) like GPT-4 possess extensive external knowledge, but research indicates that they lack specialty for NER tasks. Furthermore, non-public and large-scale weights make tuning LLMs difficult. To address these challenges, we propose a framework that combines small fine-tuned models with LLMs (LinkNER) and an uncertainty-based linking strategy called RDC that enables fine-tuned models to complement black-box LLMs, achieving better performance. We experiment with both standard NER test sets and noisy social media datasets. LinkNER enhances NER task performance, notably surpassing SOTA models in robustness tests. We also quantitatively analyze the influence of key components like uncertainty estimation methods, LLMs, and in-context learning on diverse NER tasks, offering specific web-related recommendations.
Accurate customer LifeTime Value (LTV) predictions are crucial for customer relationship management, especially in Supply Chain Platforms (SCP), which involve effectively managing the service resources in business decision-making. Previous LTV prediction methods usually rely on ample historical customer data, which is not available in the early stages of a customer's lifecycle. It makes the modeling of the historical customer data a difficult task due to the data sparsity. Besides, the long-tail distribution of customer LTV also brings new challenges to the prediction of LTV. To tackle the above issues, we propose CDLtvS, a novel Cross Domain method for customer Lifetime value prediction in SCP. It leverages rich cross-domain information from upstream platforms to enhance LTV predictions in downstream platforms. Firstly, CDLtvS pre-trains the customer representations by an LTV modeling framework named LtvS in source and target domains separately. Specifically, LtvS incorporates the Expert Mask Network (ExMN), which not only effectively models the long-tail distribution of LTV in single-domain but also resolves cross-domain learning model bias resulting from this distribution. Then, the various-level alignment mechanism is introduced to keep the consistency of knowledge transferring from source to target domains on both sparse and non-sparse data. Comprehensive experiments on real-world data from JD, one of the world's largest supply chain platforms, demonstrate that CDLtvS achieves a normalized mean average error of 0.3378 in LTV prediction, outperforming 16.3% to the baseline. Additionally, the improvements of ≥2.3% across various data sparsity levels (0% -- 80%) provide valuable insights into cross-domain LTV modeling.
In the field of online sequential decision-making, we address the problem with delays utilizing the framework of online convex optimization (OCO), where the feedback of a decision can arrive with an unknown delay. Unlike previous research that is limited to Euclidean norm and gradient information, we propose three families of delayed algorithms based on approximate solutions to handle different types of received feedback. Our proposed algorithms are versatile and applicable to universal norms. Specifically, we introduce a family of Follow the Delayed Regularized Leader algorithms for feedback with full information on the loss function, a family of Delayed Mirror Descent algorithms for feedback with gradient information on the loss function and a family of Simplified Delayed Mirror Descent algorithms for feedback with the value information of the loss function's gradients at corresponding decision points. For each type of algorithm, we provide corresponding regret bounds under cases of general convexity and relative strong convexity, respectively. We also demonstrate the efficiency of each algorithm under different norms through concrete examples. Furthermore, our theoretical results are consistent with the current best bounds when degenerated to standard settings.
The prevalence of fake news across various online sources has had a significant influence on the public. Existing Chinese fake news detection datasets are limited to news sourced solely from Weibo. However, fake news originating from multiple sources exhibits diversity in various aspects, including its content and social context. Methods trained on purely one single news source can hardly be applicable to real-world scenarios. Our pilot experiment demonstrates that the F1 score of the state-of-the-art method that learns from a large Chinese fake news detection dataset, Weibo-21, drops significantly from 0.943 to 0.470 when the test data is changed to multi-source news data, failing to identify more than one-third of the multi-source fake news. To address this limitation, we constructed the first multi-source benchmark dataset for Chinese fake news detection, termed MCFEND, which is composed of news we collected from diverse sources such as social platforms, messaging apps, and traditional online news outlets. Notably, such news has been fact-checked by 14 authoritative fact-checking agencies worldwide. In addition, various existing Chinese fake news detection methods are thoroughly evaluated on our proposed dataset in cross-source, multi-source, and unseen source ways. MCFEND, as a benchmark dataset, aims to advance Chinese fake news detection approaches in real-world scenarios.
Urban region profiling from web-sourced data is of utmost importance for urban computing. We are witnessing a blossom of LLMs for various fields, especially in multi-modal data research such as vision-language learning, where text modality serves as a supplement for images. As textual modality has rarely been introduced into modality combinations in urban region profiling, we aim to answer two fundamental questions: i) Can text modality enhance urban region profiling? ii) and if so, in what ways and which aspects? To answer the questions, we leverage the power of Large Language Models (LLMs) and introduce the first-ever LLM-enhanced framework that integrates the knowledge of text modality into urban imagery, named LLM-enhanced Urban Region Profiling with Contrastive Language-Image Pretraining (UrbanCLIP ). Specifically, it first generates a detailed textual description for each satellite image by Image-to-Text LLMs. Then, the model is trained on image-text pairs, seamlessly unifying language supervision for urban visual representation learning, jointly with contrastive loss and language modeling loss. Results on urban indicator prediction in four major metropolises show its superior performance, with an average improvement of 6.1% on R2 compared to the state-of-the-art methods. Our code and dataset are available at https://github.com/StupidBuluchacha/UrbanCLIP.
The Dialogue-level Aspect-based Sentiment Quadruple analysis (DiaASQ) task has recently received attention in the Aspect-Based Sentiment Analysis (ABSA) field. It aims to extract(target, aspect, opinion, sentiment) quadruples from multi-turn and multi-party dialogues. Compared to previous ABSA tasks focusing on text such as sentences, the DiaASQ task involves more complex contextual information and corresponding relations between terms, as well as longer sequences. These characteristics challenge existing methods that struggle to model explicit span-level interactions or have high computational costs. In this paper, we propose a span-pair interaction and tagging method to solve these issues, which includes a novel Span-pair Tagging Scheme (STS) and a simple and efficient Multi-level Representation Model (MRM). STS simplifies the DiaASQ task to a span-pair tagging task and explicitly captures complete span-level semantics by tagging span pairs. MRM efficiently models the dialogue structure information and span-level interactions by constructing multi-level contextual representation. Besides, we train a span ranker to improve the running efficiency of MRM. Extensive experiments on multilingual datasets demonstrate that our method outperforms existing state-of-the-art methods.
Dynamic pricing algorithms have been widely studied to manage hotel and platform revenue over online travel platforms (OTPs). For better dynamic pricing, the accurate estimation of the market demand and the market competitiveness are crucial. However, the existing approaches obtain a pricing strategy tailored to each specific scenario using data only from that scenario. They are not considering the shared information between different scenarios, i.e., the data from different scenarios are not fully utilized. So we propose a Multi Scenario Pricing model (MSP) with a novel sharing structure design that leverages cross-scenario and specific information to capture more accurate market demand and competitiveness. Specifically, the model structure explicitly separates information into shared components as market demand and specific information as scenario-wise price competitiveness to prevent domain seesaw. To capture the inherent correlation between listings in different scenarios, an attention network named Price Competitiveness Representation Extraction (PCRE) is well-designed. Meanwhile, traditional metrics are skewed towards model that tends to reduce the price regardless of sample distribution. Thus we propose new offline evaluation metrics that shift attention with sample distribution to avoid biased pricing strategies, which is proved to be more closely related to actual business revenue. Our proposed MSP shows superiority under both offline and online experiments on real-world datasets. The multi-scenario industry dataset and our code are available. To the best of our knowledge, it will be the first real-industry multi-scenario pricing data.
In recent years, the rise of online Knowledge Management Systems (KMSs) has significantly improved work efficiency in enterprises. Knowledge development prediction, as a critical application within these online platforms, enables organizations to proactively address knowledge gaps and align their learning initiatives with evolving job requirements. However, it still confronts challenges in exploring the influence of collaborative networks on knowledge development and adapting to ecological situations in working environment. To this end, in this paper, we propose a Collaboration-Aware Hybrid Learning approach (CAHL) for predicting the future knowledge acquisition of employees and quantifying the impact of various knowledge learning patterns. Specifically, to fully harness the inherent rules of knowledge development, we first learn the knowledge co-occurrence and prerequisite relationships with an association prompt attention mechanism to generate effective knowledge representations through a specially-designed Job Knowledge Embedding module. Then, we aggregate the features of mastering knowledge and work collaborators for employee representations in another Employee Embedding module. Moreover, we propose to model the process of employee knowledge development via a Hybrid Learning Simulation module that integrates both collaborative learning and self learning to predict future-acquired job knowledge of employees. Finally, extensive experiments conducted on a real-world dataset clearly validate the effectiveness of CAHL.
Pre-training Event Extraction (EE) models on unlabeled data is an effective strategy that frees researchers from costly and labor-intensive data annotation. However, existing pre-training methods necessitate substantial computational resources, requiring high-performance hardware infrastructure and extensive training duration. In response to these challenges, this paper proposes a Lighter, Faster, and more Data-efficient pre-training framework for EE, named LFDe. Distinct from existing methods that strive to establish a comprehensive representation space during pre-training, our framework focuses on quickly familiarizing with the task format from a small amount of automatically constructed pseudo-events. It comprises three stages: weak-label data construction, pre-training, and fine-tuning. Specifically, during the first stage, LFDe first automatically designates pseudo-triggers and arguments based on the characteristics of real events to form pre-training samples. In the processes of pre-training and fine-tuning, the framework reframes EE as the identification of tokens semantically closest to the prompt within the given sentence. This paper also introduces a novel prompt-based sequence labeling model for EE to accommodate this reframing. Experiments on real-world datasets show that compared to similar models, our framework requires fewer pre-training data (only about 0.04%), a shorter pre-training period (about 0.03%), and lower memory requirements (about 57.6%). Simultaneously, our framework significantly improves performance in various data-scarce scenarios.
While previous video to text models have achieved remarkable successes, they mostly focus on how to understand the video contents in a general sense, but fail to capture the human personalized preferences, which is highly demanded for an engaging multimodal chatbots. Different from user modeling in collaborative filtering, there is no other user behaviors in inference as a real-time video stream is coming. In this paper, we formally define the task of personalized video commenting task and design an end-to-end personalized framework for solving this task. In specific, we argue that the personalization for video comment generation can be reflected in two aspects, that is, (1) for the same video, different users may comment on different clips, and (2) for the same clip, different people may also express various opinions with diverse commentary styles. Motivated by these considerations, we design our framework based on two components. The first one is a clip selector, which is responsible for predicting the clips that the user may comment in the video. The second one is a text generator, which aims to produce the comment based on the above predicted clips and the user's preference. In our framework, these two components are optimized in an end-to-end manner to mutually enhance each other, where we design confidence-aware scheduled sampling and iterative inference strategies to solve the problem that the ground truth clips are absent in the inference phase. As the absence of personalized video to text dataset, we collect and release a new dataset for studying this problem. We conduct extensive experiments to demonstrate the effectiveness of our model.
Federated Recommendation (FedRec) systems have emerged as a solution to safeguard users' data in response to growing regulatory concerns. However, one of the major challenges in these systems lies in the communication costs that arise from the need to transmit neural network models between user devices and a central server. Prior approaches to these challenges often lead to issues such as computational overheads, model specificity constraints, and compatibility issues with secure aggregation protocols. In response, we propose a novel framework, called Correlated Low-rank Structure (CoLR), which leverages the concept of adjusting lightweight trainable parameters while keeping most parameters frozen. Our approach substantially reduces communication overheads without introducing additional computational burdens. Critically, our framework remains fully compatible with secure aggregation protocols, including the robust use of Homomorphic Encryption. The approach resulted in a reduction of up to 93.75% in payload size, with only an approximate 8% decrease in recommendation performance across datasets. Code for reproducing our experiments can be found at https://github.com/NNHieu/CoLR-FedRec.
Decentralized Collaborative Learning with Adaptive Reference Data for On-Device POI Recommendation
PDF ↗In Location-based Social Networks (LBSNs), Point-of-Interest (POI) recommendation helps users discover interesting places. There is a trend to move from the conventional cloud-based model to on-device recommendations for privacy protection and reduced server reliance. Due to the scarcity of local user-item interactions on individual devices, solely relying on local instances is not adequate. Collaborative Learning (CL) emerges to promote model sharing among users. Central to this CL paradigm is reference data, which is an intermediary that allows users to exchange their soft decisions without directly sharing their private data or parameters, ensuring privacy and benefiting from collaboration. While recent efforts have developed CL-based POI frameworks for robust and privacy-centric recommendations, they typically use a single and unified reference for all users. Reference data that proves valuable for one user might be harmful to another, given the wide range of user preferences. Some users may not offer meaningful soft decisions on items outside their interest scope. Consequently, using the same reference data for all collaborations can impede knowledge exchange and lead to sub-optimal performance. To address this gap, we introduce the Decentralized Collaborative Learning with Adaptive Reference Data (DARD) framework, which crafts adaptive reference data for effective user collaboration. It first generates a desensitized public reference data pool with transformation and probability data generation methods. For each user, the selection of adaptive reference data is executed in parallel by training loss tracking and influence function. Local models are trained with individual private data and collaboratively with the geographical and semantic neighbors. During the collaboration between two users, they exchange soft decisions based on a combined set of their adaptive reference data. Our evaluations across two real-world datasets highlight DARD's superiority in recommendation performance and addressing the scarcity of available reference data.
The heterogeneous information network (HIN), which contains rich semantics depicted by meta-paths, has emerged as a potent tool for mitigating data sparsity in recommender systems. Existing HIN-based recommender systems operate under the assumption of centralized storage and model training. However, real-world data is often distributed due to privacy concerns, leading to the semantic broken issue within HINs and consequent failures in centralized HIN-based recommendations. In this paper, we suggest the HIN is partitioned into private HINs stored on the client side and shared HINs on the server. Following this setting, we propose a federated heterogeneous graph neural network (FedHGNN) based framework, which facilitates collaborative training of a recommendation model using distributed HINs while protecting user privacy. Specifically, we first formalize the privacy definition for HIN-based federated recommendation (FedRec) in the light of differential privacy, with the goal of protecting user-item interactions within private HIN as well as users' high-order patterns from shared HINs. To recover the broken meta-path based semantics and ensure proposed privacy measures, we elaborately design a semantic-preserving user interactions publishing method, which locally perturbs user's high-order patterns and related user-item interactions for publishing. Subsequently, we introduce an HGNN model for recommendation, which conducts node- and semantic-level aggregations to capture recovered semantics. Extensive experiments on four datasets demonstrate that our model outperforms existing methods by a substantial margin (up to 34% in HR@10 and 42% in NDCG@10) under a reasonable privacy budget (e.g., ε=1).
Federated recommender systems (FedRecs) have gained significant attention for their potential to protect user's privacy by keeping user privacy data locally and only communicating model parameters/gradients to the server. Nevertheless, the currently existing architecture of FedRecs assumes that all users have the same 0-privacy budget, i.e., they do not upload any data to the server, thus overlooking those users who are less concerned about privacy and are willing to upload data to get a better recommendation service. To bridge this gap, this paper explores a user-governed data contribution federated recommendation architecture where users are free to take control of whether they share data and the proportion of data they share to the server. To this end, this paper presents a cloud-device collaborative graph neural network federated recommendation model, named CDCGNNFed. It trains user-centric ego graphs locally, and high-order graphs based on user-shared data in the server in a collaborative manner via contrastive learning. Furthermore, a graph mending strategy is utilized to predict missing links in the graph on the server, thus leveraging the capabilities of graph neural networks over high-order graphs. Extensive experiments were conducted on two public datasets, and the results demonstrate the effectiveness of the proposed method.
Click-through Rate (CTR) module is the foundation block of recommendation system and used for search, content selection, advertising, video streaming etc. CTR is modelled as a classification problem and extensive research is done to improve the CTR models. However, uncertainty method for these models are still an unexplored area. In this work we analyse popular uncertainty methods in the context of recommendation system. We found that popular uncertainty models fails to capture the predictive uncertainty of the CTR model that exist unique to the recommendation models and is not prevalent in the traditional classification models. We empirical show why a different uncertainty measure is required for the recommendation system CTR prediction models. We propose PRU (Predictive Relevance Uncertainty), a single forward pass uncertainty approach for a sample as a distance from the predictive relevance samples of the training data. We show the efficacy of the proposed predictive relevance uncertainty (PRU) on selective prediction. Further, we demonstrate the utility of the proposed framework on the downstream task of OOD detection and active learning while maintaining the latency of a single pass deterministic model.
To recommend the points of interest (POIs) that a user would check-in next, most deep-learning (DL)-based existing studies have employed random negative (RN) sampling during model training. In this paper, we claim and validate that, as the training proceeds, such an RN sampling in reality performs as sampling easy negative (EN) POIs (i.e., EN sampling) that a user was highly unlikely to check-in at her check-in time point. Furthermore, we verify that EN sampling is more disadvantageous in improving the accuracy than sampling hard negative (HN) POIs (i.e., HN sampling) that a user was highly likely to check-in. To address this limitation, we present the novel concept of the Degree of Positiveness (DoP), which can be formulated by two factors: (i) the degree to which a POI has the characteristics preferred by a user; (ii) the geographical distance between a user and a POI. Then, we propose a new model-training scheme based on HN sampling by using DoP. Using real-world datasets (i.e., NYC, TKY, and Brightkite), we demonstrate that all the state-of-the-art models trained by our scheme showed dramatic improvements in accuracy by up to about 82.8%.
Large language models (LLMs) open up new horizons for sequential recommendations, owing to their remarkable language comprehension and generation capabilities. However, there are still numerous challenges that should be addressed to successfully implement sequential recommendations empowered by LLMs. Firstly, user behavior patterns are often complex, and relying solely on one-step reasoning from LLMs may lead to incorrect or task-irrelevant responses. Secondly, the prohibitively resource requirements of LLM (e.g., ChatGPT-175B) are overwhelmingly high and impractical for real sequential recommender systems. In this paper, we propose a novel Step-by-step knowLedge dIstillation fraMework for recommendation (SLIM), paving a promising path for sequential recommenders to enjoy the exceptional reasoning capabilities of LLMs in a "slim" (i.e. resource-efficient) manner. We introduce CoT prompting based on user behavior sequences for the larger teacher model. The rationales generated by the teacher model are then utilized as labels to distill the downstream smaller student model (e.g., LLaMA2-7B). In this way, the student model acquires the step-by-step reasoning capabilities in recommendation tasks. We encode the generated rationales from the student model into a dense vector, which empowers recommendation in both ID-based and ID-agnostic scenarios. Extensive experiments demonstrate the effectiveness of SLIM over state-of-the-art baselines, and further analysis showcasing its ability to generate meaningful recommendation reasoning at affordable costs.
Graph neural networks (GNNs) have shown impressive performance in recommender systems, particularly in collaborative filtering (CF). The key lies in aggregating neighborhood information on a user-item interaction graph to enhance user/item representations. However, we have discovered that this aggregation mechanism comes with a drawback - it amplifies biases present in the interaction graph. For instance, a user's interactions with items can be driven by both unbiased true interest and various biased factors like item popularity or exposure. However, the current aggregation approach combines all information, both biased and unbiased, leading to biased representation learning. Consequently, graph-based recommenders can learn distorted views of users/items, hindering the modeling of their true preferences and generalizations. To address this issue, we introduce a novel framework called Adversarial Graph Dropout (AdvDrop). It differentiates between unbiased and biased interactions, enabling unbiased representation learning. For each user/item, AdvDrop employs adversarial learning to split the neighborhood into two views: one with bias-mitigated interactions and the other with bias-aware interactions. After view-specific aggregation, AdvDrop ensures that the bias-mitigated and bias-aware representations remain invariant, shielding them from the influence of bias. We validate AdvDrop's effectiveness on five public datasets that cover both general and specific biases, demonstrating significant improvements. Furthermore, our method exhibits meaningful separation of subgraphs and achieves unbiased representations for graph-based CF models, as revealed by in-depth analysis. Our code is publicly available at https://github.com/Arthurma71/AdvDrop/tree/main.