The rapid advancement of medical imaging technologies has led to an exponential increase in medical image data, making efficient retrieval from large-scale datasets critical for improving diagnostic accuracy and speed. However, two key challenges hinder this process: first, the presence of uncertain and subtle lesions in medical images that are often difficult to discern, and second, class imbalance across different case types within medical image databases. These inherent challenges significantly degrade the performance of existing hashing algorithms. In recent years, methods based on the Takagi–Sugeno–Kang fuzzy system (TSK-FS) have shown promising performance in medical image modeling. Inspired by these advances, this article proposes a novel fuzzy hashing network (FHN) based on TSK-FS to enhance retrieval performance by effectively handling both uncertainty and data imbalance in medical imaging. The FHN first introduces a novel fuzzification mechanism that incorporates the concept of a self-attention mechanism to effectively capture the complex underlying features in medical images, thereby enhancing the data discriminability in fuzzy spaces. Meanwhile, a new consequent parameter learning mechanism is developed for defuzzification by introducing the Transformer network, which aims to improve the inference efficiency and generalization capability of the FHN. Based on these two mechanisms, FHN's capability of analyzing and handling uncertain data is significantly enhanced. Furthermore, a novel hash center loss is designed to capture global relationships while emphasizing local structural information, thereby improving the handling of imbalanced data and significantly enhancing retrieval performance.

Probabilistic queries usually suffer from the noisy query result sets, due to data uncertainty. In this paper, we propose an efficient optimization framework, termed as QueryClean, for both probabilistic skyline computation and probabilistic similarity search. Its goal is to optimize query quality by selecting a group of uncertain objects to clean under limited resource available, where an entropy based quality function is leveraged. We develop an efficient index to organize the possible result sets of probabilistic queries, which is able to help avoid multiple probabilistic query evaluations over a large number of possible worlds for quality computation. Moreover, using two newly presented heuristics, we present exact and approximate algorithms for the optimization problem. Extensive experiments on both real and synthetic data sets demonstrate the efficiency and scalability of QueryClean.

Probabilistic queries have been extensively explored to provide answers with confidence, in order to support the real-life applications struggling with uncertain data, such as sensor networks and data integration. However, the uncertainty of data may propagate, and thus, the results returned by probabilistic queries contain much noise, which degrades query quality significantly. In this paper, we propose an efficient optimization framework, termed as QueryClean, for both probabilistic skyline computation and probabilistic similarity search. The goal of QueryClean is to optimize query quality via selecting a group of uncertain objects to clean under limited resource available, where a joint-entropy based quality function is leveraged. We develop an efficient structure called ASI to index the possible result sets of probabilistic queries, which helps to avoid many types of probabilistic query evaluations over a large number of the possible worlds for quality computation. Moreover, we present exact and approximate algorithms for the optimization problem, using two newly presented heuristics. Considerable experimental results on both real and synthetic data sets demonstrate the efficiency and scalability of our proposed framework QueryClean.

This paper explores the causality and responsibility problem (CRP) for the non-answers to probabilistic reverse skyline queries (PRSQ). Towards this, we propose an efficient algorithm called CP to compute the causality and responsibility for the non-answers to PRSQ. CP first finds candidate causes, and then, it performs verification to obtain actual causes with their responsibilities, during which several strategies are used to boost efficiency. Extensive experiments using both real and synthetic data sets demonstrate the effectiveness and efficiency of the presented algorithms.

We develop IS2R, an efficient interactive system for refining reverse top-k queries, to eliminate unexpected query results including (i) the absence of expected objects, (ii) the presence of unexpected objects, and (iii) the empty query result. The IS2R returns the refinement suggestions with the minimal costs based on penalty models. In this demonstration (available at https://youtu.be/GnSm4T9Uslk), we show different scenarios on how IS2R can be used to refine the original reverse top-k query, and verify its effectiveness and efficiency.

Causality and responsibility is an essential tool in the database community for providing intuitive explanations for answers/non-answers to queries. Causality denotes the causes for the answers/non-answers to queries, and responsibility represents the degree of a cause which reflects its influence on the answers/non-answers to queries. In this paper, we study the causality and responsibility problem (CRP) for the non-answers to probabilistic reverse skyline queries (PRSQ). We first formalize CRP on PRSQ, and then, we propose an efficient algorithm termed as CP to compute the causality and responsibility for the non-answers to PRSQ. CP first finds candidate causes, and then, it performs verification to obtain actual causes with their responsibilities, during which several strategies are used to boost efficiency. Further, we explore the CRP for the non-answers to reverse skyline queries. Towards this, we extend CP to identify directly all the actual causes and their responsibilities for a non-answer to reverse skyline queries without additional verification. Extensive experiments using both real and synthetic data sets demonstrate the effectiveness and efficiency of our presented algorithms.

Why-not questions, which aim to seek clarifications on the missing tuples for query results, have recently received considerable attention from the database community. In this paper, we systematically explore why-not questions on reverse top-k queries , owing to its importance in multi-criteria decision making. Given an initial reverse top- k query and a missing/why-not weighting vector set W m that is absent from the query result, why-not questions on reverse top- k queries explain why W m does not appear in the query result and provide suggestions on how to refine the initial query with minimum penalty to include W m in the refined query result. We first formalize why-not questions on reverse top- k queries and reveal their semantics, and then propose a unified framework called WQRTQ to answer why-not questions on both monochromatic and bichromatic reverse top- k queries. Our framework offers three solutions, namely, (i) modifying a query point q , (ii) modifying a why-not weighting vector set W m and a parameter k , and (iii) modifying q , W m , and k simultaneously, to cater for different application scenarios. Extensive experimental evaluation using both real and synthetic data sets verifies the effectiveness and efficiency of the presented algorithms.