The effectiveness upper bound of retrieval-augmented generation (RAG) is fundamentally constrained by the semantic integrity and information granularity of text chunks in its knowledge base. Moreover, domain documents are characterized by dense terminology and strong contextual dependencies, which exacerbate the semantic fragmentation of text chunks, thereby making it difficult to efficiently utilize their key information. To address these challenges, this paper proposes QChunker, which restructures the RAG paradigm from retrieval-augmentation to understanding-retrieval-augmentation. Firstly, QChunker models the text chunking as a composite task of text segmentation and knowledge completion to ensure the logical coherence and integrity of text chunks. Drawing inspiration from Hal Gregersen's ''Questions Are the Answer'' theory, we design a multi-agent debate framework comprising four specialized components: a question outline generator, text segmenter, integrity reviewer, and knowledge completer. This framework operates on the principle that questions serve as catalysts for profound insights. Through this pipeline, we successfully construct a high-quality dataset of 45K entries and transfer this capability to small language models. Additionally, to handle long evaluation chains and low efficiency in existing chunking evaluation methods, which overly rely on downstream QA tasks, we introduce a novel direct evaluation metric, ChunkScore. Both theoretical and experimental validations demonstrate that ChunkScore can directly and efficiently discriminate the quality of text chunks. Furthermore, during the text segmentation phase, we utilize document outlines for multi-path sampling to generate multiple candidate chunks and select the optimal solution employing ChunkScore. Extensive experimental results across four heterogeneous domains exhibit that QChunker effectively resolves aforementioned issues by providing RAG with more logically coherent and information-rich text chunks. Notably, this study also establishes a small-domain QA dataset concerning hazardous chemical safety, which fully reveals the significant value of RAG in specialized domains and the generalization capability of the QChunker framework.

Multi-modal entity alignment (MMEA) aims to find the equivalent entities between multi-modal knowledge graphs (MMKGs). Current MMEA methods follow the ''embed-fuse-compare'' paradigm and show decent performance improvements on several public datasets. However, this paradigm may fail to fully address inconsistencies across different modalities, resulting in instability and sub-optimal performance. In this paper, we propose a novel paradigm called ''embed-fuse-assign'', which transforms MMEA into a multi-modal assignment problem. First, we prove that multi-modal entity alignment can be turned into assignment problems and identify that obtaining correspondence scores is the crucial step. Then, we devise a novel Explicit and Implicit multi-modal Entity Alignment (EIEA) algorithm, which classifies modalities into explicit and implicit categories based on the need for joint neighborhood information, designs modality-specific embedding and correlation scoring mechanisms, and derives alignment results by integrating correspondences across all modalities. Finally, we introduce a conflict-aware soft pseudo-labeling method to further optimize semi-supervised learning for implicit modality. Extensive experiments have demonstrated that our proposed paradigm and algorithm achieve the state-of-the-art performance on five real-world MMEA datasets. The source code and datasets are released at https://github.com/Bubble-bubble77/EIEA.

Display advertising plays a crucial role in benefiting advertisers, publishers, and users. Traditional display advertising systems employ a multi-stage architecture comprising retrieval, coarse ranking, ranking, and re-ranking. However, conventional retrieval methods primarily rely on ID-based learning-to-rank mechanisms, often underutilizing the content information of ads, like ads' title, and description. This limitation reduces the ability to generate diverse and relevant recommendation lists. To address this challenge, we propose leveraging the extensive world knowledge of large language models (LLMs). However, effectively integrating LLMs into advertising systems presents three key challenges: ( i) How to accurately capture user interests, (ii) How to bridge the knowledge gap between LLMs and advertising systems , and ( iii) How to efficiently deploy LLMs at scale. To overcome these challenges, we introduce LEADRE —the L LM E mpowered Display AD vertisement RE commender system. LEADRE consists of three core components. The Intent-Aware Prompt Engineering module introduces multi-faceted knowledge and constructs intent-aware pairs, fine-tuning LLMs to generate ads tailored to users' personal interests. The Advertising-Specific Knowledge Alignment module incorporates auxiliary fine-tuning tasks and Direct Preference Optimization (DPO) to align LLMs with advertising semantics and business objectives. The Latency-Aware Model Deployment module integrates a hybrid service framework that balances latency-tolerant and latency-sensitive service, ensuring seamless online deployment. Extensive offline experiments validate the effectiveness of LEADRE, demonstrating significant improvements across multiple evaluation metrics. Furthermore, online A/B tests reveal a 1.57% and 1.17% increase in Gross Merchandise Value (GMV) for serviced users on WeChat Channels and Moments, respectively. LEADRE has been successfully deployed on both platforms, handling tens of billions of requests daily.

Recently, Grounded Multimodal Named Entity Recognition (GM-NER) task has been introduced to refine the Multimodal Named Entity Recognition (MNER) task.Existing MNER studies fall short in that they merely focus on extracting text-based entity-type pairs, often leading to entity ambiguities and failing to contribute to multimodal knowledge graph construction.In the GMNER task, the objective becomes more challenging: identifying named entities in text, determining their entity types, and locating their corresponding bounding boxes in linked images, necessitating precise alignment between the textual and visual information.We introduce a novel multi-level alignment pre-training method, engaging with both text-image and entity-object dimensions to foster deeper congruence between multimodal data.Specifically, we innovatively harness potential objects identified within images, aligning them with textual entity prompts, thereby generating refined soft pseudolabels.These labels serve as self-supervised signals that pre-train the model to more accurately extract entities from textual input.To address misalignments that often plague modality integration, our method employs a sophisticated diffusion model that performs back-translation on the text to generate a corresponding visual representation, thus refining the model's multimodal interpretative accuracy.Empirical evidence from the GMNER dataset validates that our approach significantly outperforms existing state-of-theart models.Moreover, the versatility of our pre-training process complements virtually all extant models, offering an additional avenue for augmenting their multimodal entity recognition acumen.

Multimodal named entity recognition (MNER) is a vision-language task, which aims to detect entity spans and classify them to corresponding entity types given a sentence-image pair. Existing methods often regard an image as a set of visual objects, trying to explicitly capture the relations between visual objects and entities. However, since visual objects are often not identical to entities in quantity and type, they may suffer the bias introduced by visual objects rather than aid. Inspired by the success of textual prompt-based fine-tuning (PF) approaches in many methods, in this paper, we propose a Multimodal Prompt-based Machine Reading Comprehension based framework to implicit alignment between text and image for improving MNER, namely MPMRC-MNER. Specifically, we transform text-only query in MRC into multimodal prompt containing image tokens and text tokens. To better integrate image tokens and text tokens, we design a prompt-aware attention mechanism for better cross-modal fusion. At last, contrastive learning with two types of contrastive losses is designed to learn more consistent representation of two modalities and reduce noise. Extensive experiments and analyses on two public MNER datasets, Twitter2015 and Twitter2017, demonstrate the better performance of our model against the state-of-the-art methods.

In this paper, we study inequality query (IQ query) processing in tuple independent probabilistic databases, where IQ queries can be categorized into IQ-path, IQ-tree, and IQ-graph queries. We focus on two related issues for IQ queries. One issue is to efficiently compute their probabilities, with the observation that the time complexity of the state-of-the-art algorithm to process IQ-graph queries is high. The other issue is to efficiently perform their sensitivity analysis, which has not been studied before. Here, sensitivity analysis is to identify input tuples that have high influence on the probability of an answer tuple, and the influence of an input tuple is defined as the difference between the output probabilities obtained in two cases, where we assume that the tuple exists in one case and does not exist in the other one. In this paper, we compile the inequality conditions of an IQ query q into a compilation tree T, which encodes the Shannon expansion order. Moreover, we split q into a set of subqueries and each contains only one inequality condition. Using compilation tree and decomposition, we introduce a dynamic programming algorithm called Dec to process an IQ query q in time O(IΦI), where Φ is the lineage of q. An IQ query can be processed by our Decif and only if its inequality conditions can be compiled into a compilation tree T and the inequality conditions from any node to all of its child nodes must be the same in T. We conduct extensive experiments using real and synthetic datasets to demonstrate the efficiency of our algorithm for computing the probabilities and influences of IQ queries.

This paper investigates the problem of efficiently computing responsibility for lineages of conjunctive queries with inequalities on databases. We classify the lineages of a class of queries with inequalities, called IQ queries, into path and composite lineages. We first compile path lineages into lineage graphs and transform lineage graphs into matrices. Then we reduce the problem of computing responsibility for path lineages to the shortest path problem, which can be solved by the dynamic programming algorithm in PTIME. We further prove composite lineages can be decomposed into path lineages for computing responsibility. Thus, our first main result shows it is in PTIME to compute responsibility for lineages of IQ queries. We generalize the previous results on dichotomy of responsibility analysis for lineages of conjunctive queries with equalities, now in the presence of inequalities. After decomposing composite lineages into path lineages, the data population needed for computing responsibility decreases more than one order of magnitude. Thus, our algorithm can efficiently compute responsibility for composite lineages. In order to compute responsibility for lineages in general, we introduce a greedy algorithm, consisting of a reduction to the set cover problem. Finally, we demonstrate the benefits of the proposed algorithms with extensive experimental results.

This article presents a set of artificial surfaces simulating complex terrain types for evaluating the performances of grid-based flow direction algorithms. The proposed artificial surfaces were developed based on sine and cosine functions and thus can simulate four complex terrain types: a convex-centred slope, concave-centred slope, saddle-centred slope and straight-ridge slope; such features are typical and widespread in real landscapes. We analytically solved the theoretical values of specific catchment area (SCA) for the proposed artificial surfaces. Compared with existing artificial surfaces for evaluating flow direction algorithms, the proposed artificial surfaces provide a better representation of common terrain types in the real world. To analyse the feasibility of the proposed artificial surfaces, two sets of artificial digital elevation models (DEMs) were created by sampling the proposed artificial surfaces with different reliefs at a series of resolutions (i.e. 1, 5, 10 and 20 m). Four representative flow direction algorithms were applied to these artificial DEMs: D8, D-inf, FD8 and MFD-md. The root mean square error, mean error and standard deviation in the computed SCA from flow direction algorithm show that MFD-md generally yielded lower error under simulated terrain conditions than D8, D-inf and FD8. The cumulative frequency distributions of errors from the tested flow direction algorithms with the proposed artificial DEMs can effectively reflect the inherent characteristics of each algorithm. MFD-md performed more similarly to FD8 in low-relief terrains and more similarly to D-inf in high-relief terrains. The map of errors from each tested algorithm is available for a spatially explicit evaluation of the occurrence of errors. Over most of the area, the D-inf algorithm underestimated the SCA when FD8 overestimated the SCA.

Data uncertainty is common in real-world applications due to various causes, including imprecise measurement, network latency, outdated sources and sampling errors. These kinds of uncertainty have to be handled cautiously, or else the mining results could be unreliable or even wrong. In this paper, we propose a new rule-based classification and prediction algorithm called uRule for classifying uncertain data. This algorithm introduces new measures for generating, pruning and optimizing rules. These new measures are computed considering uncertain data interval and probability distribution function. Based on the new measures, the optimal splitting attribute and splitting value can be identified and used for classification and prediction. The proposed uRule algorithm can process uncertainty in both numerical and categorical data. Our experimental results show that uRule has excellent performance even when data is highly uncertain.