Similarity Queries Research Articles

Effective Hard Negative Mining for Contrastive Learning-based Code Search

Background. Code search aims to find the most relevant code snippet in a large codebase based on a given natural language query. An accurate code search engine can increase code reuse and improve programming efficiency. The focus of code search is how to represent the semantic similarity of code and query. With the development of code pre-trained models, the pattern of using numeric feature vectors (embeddings) to represent code semantics and using vector distance to represent semantic similarity has replaced traditional string matching methods. The quality of semantic representations is critical to the effectiveness of downstream tasks such as code search. Currently, the state-of-the-art (SOTA) learning method uses the contrastive learning paradigm. The objective of contrastive learning is to maximize the similarity between matching code and query (positive samples) and minimize the similarity between mismatched pairs (negative samples). To increase the reusing of negative samples, prior contrastive learning approaches use a large queue (memory bank) to store embeddings. Problem. However, there is still a lot of room for improvement in using negative examples for code search: ① Due to the random selection of negative samples, semantic representations learned by existing models cannot distinguish similar codes well. ② Since semantic vectors in the memory bank are reused from previous inference results and then directly used for loss function calculation without gradient descent, the model cannot effectively learn the negative sample semantic information. Method. To solve the above problems, we propose a contrastive learning code search model with hard negative mining called CoCoHaNeRe: ❶ To enable the model to distinguish similar codes, we introduce hard negative examples into contrastive training, which are negative examples in the codebase that are most similar to positive examples. As a result, hard negative examples are most likely to make the model make mistakes. ❷ To improve the learning efficiency of negative samples during training, we add all hard negative examples to the model's gradient descent process. Result. To verify the effectiveness of CoCoHaNeRe, we conducted experiments on large code search datasets with six programming languages, as well as similar retrieval tasks code clone detection and code question answering. Experimental results show that our model achieves SOTA performance. In the code search task, the average MRR score of CoCoHaNeRe exceeds CodeBERT, GraphCodeBERT, and UniXcoder by 11.25%, 8.13%, and 7.38%, respectively. It has also made great progress in code clone detection and code question answering. In addition, our method performs well in different programming languages and code pre-training models. Furthermore, qualitative analysis shows that our model effectively distinguishes high-order semantic differences between similar codes.

Caching Historical Embeddings in Conversational Search

Rapid response, namely, low latency, is fundamental in search applications; it is particularly so in interactive search sessions, such as those encountered in conversational settings. An observation with a potential to reduce latency asserts that conversational queries exhibit a temporal locality in the lists of documents retrieved. Motivated by this observation, we propose and evaluate a client-side document embedding cache, improving the responsiveness of conversational search systems. By leveraging state-of-the-art dense retrieval models to abstract document and query semantics, we cache the embeddings of documents retrieved for a topic introduced in the conversation, as they are likely relevant to successive queries. Our document embedding cache implements an efficient metric index, answering nearest-neighbor similarity queries by estimating the approximate result sets returned. We demonstrate the efficiency achieved using our cache via reproducible experiments based on Text Retrieval Conference Conversational Assistant Track datasets, achieving a hit rate of up to 75% without degrading answer quality. Our achieved high cache hit rates significantly improve the responsiveness of conversational systems while likewise reducing the number of queries managed on the search back-end.

Enhanced query performance for stored streaming data through structured streaming within spark SQL

Traditional database systems like relational databases can store data which are structured with predefined schema, but in the case of bigdata, the data comes in different formats or are collected from diverse sources. The distributed databases like not only spark querying language (NoSQL) repositories are often used in relation to bigdata analytics, but a continual updating is required in business because of the streaming data that comes from stock trading, online activities of website visitors, and from the mobile applications in real time. It will not have to delay, for some report to show up, to assess and analyse the current situation, to move forward with the next business choice. Apache Spark’s structured streaming offer capabilities for handling streaming data in a batch processing mode with faster responses compared to MongoDB which is a document-based NoSQL database. This study completes similar queries to evaluate Spark SQL and NoSQL database performance, focusing on the upsides of Spark SQL over NoSQL databases in streaming data exploration. The queries are completed with streaming data stored in a batch mode.

Secure similar patients query with homomorphically evaluated thresholds

Patient-centric precision medicine requires the analysis of large volumes of genomic data to tailor treatments and medications based on individual-level characteristics. Because the amount of data held by a single institution is limited, researchers may want access to genomic data held by other institutions. Owing to the inherent privacy implications of genomic data, performing comparisons on encrypted data is preferable in certain settings. The Similar patient query (SPQ) is an application that enables a secure search across genomic databases for patients with similar genetic makeup. Query results can be used to draw meaningful conclusions regarding suitable therapies.However, existing protocols either reveal intermediate computations, such as similarity scores, which can lead to membership-inference attacks, or they realize the ideal Boolean output (similar/not similar) through multiple protocol rounds, requiring the database owners to stay online throughout.This paper introduces a two-party privacy-preserving approach to perform SPQs across encrypted genomic databases based on secure function extensions of additively homomorphic encryption. In contrast to related works, our scheme enables secure computation of genomic data similarity without an external party in a single round. This is achieved for more than 1000 positions of a genome in a single public key operation of 256-bit security level in the integer factorization setting.

Presto's History-Based Query Optimizer

An important feature of modern query optimizers is the ability to produce a query plan that is optimal for the underlying data set. This requires the ability to estimate cardinalities and computational costs of intermediate query plan nodes, which is highly dependent on both the query shape and the underlying data distribution. Traditional methods include collecting statistics on base tables and implementing cardinality and computational cost derivation inside the optimizer, which is error-prone for complex query shapes. This paper presents Presto's novel history-based optimization framework (HBO), which collects execution histories and uses them to optimize similar queries in the future. The framework produces accurate estimates for complex query shapes in a lightweight, automated manner, and adapts automatically to changes in underlying data distributions. We present the design and implementation of the HBO framework and provide details on its use in various optimization rules, as well as details on implementing the statistics store on top of a Redis key-value store. We also present the results of running HBO in production in two large data infrastructure organizations (Meta and Uber).

FedSQ: A Secure System for Federated Vector Similarity Queries

Vector databases have emerged as crucial tools for managing and retrieving representation embeddings of unstructured data. Given the explosive growth of data, vector data is often distributed and stored across multiple organizations. However, privacy concerns and regulations like GDPR present new challenges in collaborative and secure queries, also known as federated queries, over those vector data distributed across various data owners. Although existing research has attempted to enable such query services for low-dimensional data, such as relational and spatial data, these solutions can be inefficient in answering vector similarity queries involving high-dimensional data. Therefore, we are motivated to develop a new prototype system called FedSQ that (1) ensures privacy protection across data owners and (2) balances query efficiency and result accuracy when processing federated vector similarity queries. To achieve these goals, FedSQ utilizes advanced secure multi-party computation techniques to prevent information leakage during query processing and incorporates indexing and sampling based optimizations to strike a proper performance balance.

Does generative artificial intelligence pose a risk to performance validity test security?

Objective We examined the performance validity test (PVT) security risk presented by artificial intelligence (AI) chatbots asking questions about neuropsychological evaluation and PVTs on two popular generative AI sites. Method In 2023 and 2024, multiple questions were posed to ChatGPT-3 and Bard (now Gemini). One set started generally and refined follow-up questions based on AI responses. A second set asked how to feign, fake, or cheat. Responses were aggregated and independently rated for inaccuracy and threat. Responses not identified as inaccurate were assigned a four-level threat rating (no, mild, moderate, or high threat). Combined inaccuracy and threat ratings were examined cross-sectionally and longitudinally. Results Combined inaccuracy rating percentages were 35 to 42% in 2023 and 16 to 28% in 2024. Combined moderate/high threat ratings were observed in 24 to 41% of responses in 2023 and in 17 to 31% of responses in 2024. More ChatGPT-3 responses were rated moderate or high threat compared to Bard/Gemini responses. Over time, ChatGPT-3 responses became more accurate with a similar threat level, but Bard/Gemini responses did not change in accuracy or threat. Responses to how to feign queries demonstrated ethical opposition to feigning. Responses to similar queries in 2024 showed even stronger ethical opposition. Conclusions AI chatbots are a threat to PVT test security. A proportion of responses were rated as moderate or high threat. Although ethical opposition to feigning guidance increased over time, the natural language interface and the volume of AI chatbot responses represent a potentially greater threat than traditional search engines.

Mitochondrial Genome Resource of the Cottony Ash Psyllid, a Host of a Newly Identified ‘Candidatus Liberibacter’ Bacterium

Cottony ash psyllid (CAP, Psyllopsis discrepans) is an important, invasive insect pest of ash trees in North America, where it has established populations and is the host of a newly identified strain of ‘ Candidatus Liberibacter solanacearum’. However, not much is known about the diversity of its introduced population. In this study, a CAP mitochondrial genome (mitogenome) sequence was obtained from a collection in Saskatoon, Saskatchewan, Canada. The CAP mitogenome is a circular DNA of 18,824 bp, encoding 13 protein-coding genes, 21 transfer RNA genes, and two ribosomal RNA genes. BLAST search using the CAP mitogenome as a query against the GenBank sequence database showed the mitogenome of Euphyllura phillyreae (15,202 bp) was the most similar (query coverage = 77%; percentage identity = 78.90%). The CAP mitogenome is significantly different from other known psyllid mitogenomes with the presence of a 4,357-bp control region. The mitogenome sequence will further the genomic understanding of CAP. [Formula: see text] The author(s) have dedicated the work to the public domain under the Creative Commons CC0 “No Rights Reserved” license by waiving all of his or her rights to the work worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law, 2024.

A privacy-preserving image retrieval scheme with access control based on searchable encryption in media cloud

With the popularity of the media cloud computing industry, individuals and organizations outsource image computation and storage to the media cloud server to reduce the storage burden. Media images usually contain a large amount of private information. To prevent disclosure of privacy of the image owners, media images are encrypted before uploading to the server. However, this operation will greatly limit the utilization of the image for the user, such as content-based image retrieval. We propose an efficient similarity query algorithm with access control based on Bkd-tree in this paper, in which a searchable encryption scheme is designed for similarity image retrieval, and the encrypted image is used to extract image features by a pre-trained CNN model. The Bkd-tree is utilized to generate an index tree for the image features to speed up retrieval and make it faster than linear indexing. Finally, the security performances of the proposed scheme is analyzed and the performance of this scheme is evaluated by experiments. The results show that the security of the image content and image features can be ensured, and it has a shorter retrieval time and higher retrieval efficiency.

Exploiting Shared Sub-Expression and Materialized View Reuse for Multi-Query Optimization

AbstractQuerying in isolation lacks the potential of reusing intermediate results, which ends up wasting computational resources. Multi-Query Optimization (MQO) addresses this challenge by devising a shared execution strategy across queries, with two generally used strategies: batched or cached. These strategies are shown to improve performance, but hardly any study explores the combination of both. In this work we explore such a hybrid MQO, combining batching (Shared Sub-Expression) and caching (Materialized View Reuse) techniques. Our hybrid-MQO system merges batched query results as well as caches the intermediate results, thereby any new query is given a path within the previous plan as well as reusing the results. Since caching is a key component for improving performance, we measure the impact of common caching techniques such as FIFO, LRU, MRU and LFU. Our results show LRU to be the optimal for our usecase, which we use in our subsequent evaluations. To study the influence of batching, we vary the factor - - which represents the similarity of the results within a query batch. Similarly, we vary the cache sizes to study the influence of caching. Moreover, we also study the role of different database operators in the performance of our hybrid system. The results suggest that, depending on the individual operators, our hybrid method gains a speed-up between 4x to a slowdown of 2x from using MQO techniques in isolation. Furthermore, our results show that workloads with a generously sized cache that contain similar queries benefit from using our hybrid method, with an observed speed-up of 2x over sequential execution in the best case.

Optimization study of pruning strategy based on KNN trajectory similarity query

With the development of GPS positioning technology, a large amount of spatiotemporal trajectory data has been generated. Due to the complex structure of trajectory data, which has irregular spatial shapes and continuous temporal sequence attributes, querying massive trajectory data poses certain challenges. The pruning strategy of existing trajectory similarity query methods is not very effective. Even after pruning operations, there are still a large number of trajectories that need to undergo distance calculations to confirm whether they are similar trajectories. This paper proposes multiple local pruning optimization schemes, which maximally reduce the number of trajectories in the candidate set. Specifically, it starts with region pruning in the index space, eliminating index spaces with distances greater than the maximum similarity distance from the query trajectory. Then, it performs distance pruning between trajectories, removing trajectories that do not meet the distance conditions. Finally, it adds a termination condition: when the distance between the current index space and the query trajectory is greater than the maximum similarity distance and the number of trajectories in the result set is K, the loop is exited, and the query process ends. Tests on real datasets demonstrate that the KTSS method outperforms current algorithms of the same type.

Vexless: A Serverless Vector Data Management System Using Cloud Functions

Cloud functions, exemplified by AWS Lambda and Azure Functions, are emerging as a new computing paradigm in the cloud. They provide elastic, serverless, and low-cost cloud computing, making them highly suitable for bursty and sparse workloads, which are quite common in practice. Thus, there is a new trend in designing data systems that leverage cloud functions. In this paper, we focus on vector databases, which have recently gained significant attention partly due to large language models. In particular, we investigate how to use cloud functions to build high-performance and cost-efficient vector databases. This presents significant challenges in terms of how to perform sharding, how to reduce communication overhead, and how to minimize cold-start times. In this paper, we introduce Vexless, the first vector database system optimized for cloud functions. We present three optimizations to address the challenges. To perform sharding, we propose a global coordinator (orchestrator) that assigns workloads to Cloud function instances based on their available hardware resources. To overcome communication overhead, we propose the use of stateful cloud functions, eliminating the need for costly communications during synchronization. To minimize cold-start overhead, we introduce a workload-aware Cloud function lifetime management strategy. Vexless has been implemented using Azure Functions. Experimental results demonstrate that Vexless can significantly reduce costs, especially on bursty and sparse workloads, compared to cloud VM instances, while achieving similar or higher query performance and accuracy.

Molecular characterization of pathogenic Aeromonas jandaei bacteria isolated from cultured walking catfish (Clarias sp.)

Abstract. Mulia DS, Dwi NR, Suwarsito, Muslimin B. 2024. Molecular characterization of pathogenic Aeromonas jandaei bacteria isolated from cultured walking catfish (Clarias sp.). Biodiversitas 25: 1185-1193. Aeromonas spp. is pathogenic bacteria capable of causing fish disease, including walking catfish (Clarias sp.). In bacteria, virulence genes play a significant role in determining pathogenicity, while resistance genes influence resistivity. Therefore, this study aimed to determine molecular characterization of pathogenic Aeromonas jandaei bacteria isolated from walking catfish. The purposive sampling method was used to obtain diseased walking catfish from aquaculture ponds in Banyumas District, Central Java Province, Indonesia. The identification of bacteria was molecularly carried out based on 16S rDNA, followed by the detection of virulence genes on aer/haem, alt, ast, flaA, lafA, and fstA, as well as resistance genes on tet-A, strA-strB, and qnrA. The results of molecular analysis showed that three isolates from Singasari, namely BmSL-02, BmSL-04, and BmSL-07, were closely related to A. jandaei strain CDC0787-80 with a similarity rate and query coverage of 99.86% and 99%, respectively. Meanwhile, isolates from Cikawung, including BmCL-02, BmCL-05, and BmCL-07 were closely related to A. jandaei strain ATCC 49568. A total of six A. jandaei isolates had aer/haem virulence genes while alt, ast, and fstA were not found. BmSL-04 and BmCL-07 isolates had flaA gene, while four isolates including BmSL-02, BmSL-04, BmSL-07, and BmCL-07 had lafA gene. Additionally, BmCL-05 had tetA resistance gene, while the strA-strB and qnrA genes were not detected in all isolates. A. jandaei isolates were resistant to bacitracin, tetracycline, and gentamycin, but BmSL-04, BmSL-07, BmCL-02, and BmCL-05 were sensitive to chloramphenicol. This study provided valuable information on molecular characteristics of A. jandaei, serving as guidance in effective control and monitoring of isolates from Indonesia, specifically Banyumas, Central Java.

Secure DNA Data Storage with Similarity Search in Cloud Environments

DNA similarity search has been widely applied in human genomic studies including DNA marking, genomic sequencing and genetic disease prediction. Meanwhile, with the explosive growth of data, users are increasingly inclining to store DNA data on the cloud for saving local cost. However, the high sensitivity of DNA data has forced the government to strictly control its acquisition and utilization. One potential solution is to encrypt DNA data before outsourcing them to the cloud. Nevertheless, private DNA similarity query has been an active research issue, state-of-the-art results are still defective in security, functionality and efficiency. In this paper, we propose EFSS, an Efficient and Fine-grained Similarity Search scheme over encrypted DNA data. In specific, firstly, we design an approximation algorithm to efficiently calculate the edit distances between two sequences. Secondly, we put forward a novel Boolean search strategy to achieve complicated logic queries such as mixed “AND” and “NO” operations on genes. Thirdly, data access control is also supported in our EFSS through a variant of polynomial based design. Moreover, the K-means clustering algorithm is exploited to further improve the efficiency of execution. In the end, security analysis and extensive experiments demonstrate the high performance of EFSS compared with existing schemes.

A LITERATURE REVIEW: THE IMPORTANCE OF TERM NORMALIZATION IN VECTOR SPACE MODEL

In today's digitally inundated era, accessing information is more accessible yet challenging due to the sheer volume available. This article underscores the pivotal role of VSM in managing vast data and enhancing retrieval accuracy by ranking documents based on query similarity. Term normalization, a part of VSM development, standardizes words for indexing, improving accuracy by addressing word variations. The study's methodology involved a systematic literature review, data collection via electronic databases, and thematic analysis. The research findings highlight vital aspects: the fundamentals of information retrieval systems, the working principle of VSM in document sorting, and the process of term normalization. Various methods within term normalization, such as tokenizing, filtering, stemming, and term weighting (e.g., TF, IDF, Cosine Similarity), are elucidated for refining document relevance. Discussions underscore the impact of term normalization on information retrieval, emphasizing heightened accuracy, efficiency, and reduced error rates. In the research paper, five studies that showcased successful applications of VSM across diverse domains were referenced. These domains included karaoke song searches, thesis examiner selection, pest identification in rice plants, hadith interpretation, and library material searches. Each study demonstrated the effectiveness and versatility of VSM in solving various problems in different fields. In conclusion, VSM emerges as a potent tool in managing information overload, particularly when coupled with normalization techniques. The studies reviewed illustrate VSM's efficacy in delivering precise results, affirming its status as a preferred method in information retrieval systems due to its accuracy and effectiveness.

QuerySum: A Multi-Document Query-Focused Summarization Dataset Augmented with Similar Query Clusters

Query-focused summarization (QFS) aims to summarize the source document(s) with regard to a specific aspect of information given in a query. It plays an important role in presenting users with a concise answer summary from a set of query-relevant documents retrieved by the information retrieval system. Nonetheless, the QFS research has long been hampered by the lack of adequate datasets in terms of both quality and quantity. In this paper, we introduce a large-scale multi-document query-focused summarization dataset, called QuerySum, which contains 27,041 data samples covering diverse topics and its quality is guaranteed through human verification. Unlike some previous QFS datasets constructed directly from the question answering datasets, 74% queries in our dataset are the challenging non-factoid What-, Why-, and How- questions. More importantly, we also provide a set of similar queries together with the corresponding summaries pairs for each query as the retrieved context, presenting a new feature of QuerySum. We aim to encourage research efforts in query intention understanding in the context of QFS. Leveraging QuerySum's depth, we propose a model for query-aware multi-document summarization and set a new QFS benchmark.

KMSQ: Efficient and Privacy-Preserving Keyword-Oriented Multidimensional Similarity Query in eHealthcare

KMSQ: Efficient and Privacy-Preserving Keyword-Oriented Multidimensional Similarity Query in eHealthcare

Spatial-Temporal Graph Enhanced DETR Towards Multi-Frame 3D Object Detection.

The Detection Transformer (DETR) has revolutionized the design of CNN-based object detection systems, showcasing impressive performance. However, its potential in the domain of multi-frame 3D object detection remains largely unexplored. In this paper, we present STEMD, a novel end-to-end framework that enhances the DETR-like paradigm for multi-frame 3D object detection by addressing three key aspects specifically tailored for this task. First, to model the inter-object spatial interaction and complex temporal dependencies, we introduce the spatial-temporal graph attention network, which represents queries as nodes in a graph and enables effective modeling of object interactions within a social context. To solve the problem of missing hard cases in the proposed output of the encoder in the current frame, we incorporate the output of the previous frame to initialize the query input of the decoder. Finally, it poses a challenge for the network to distinguish between the positive query and other highly similar queries that are not the best match. And similar queries are insufficiently suppressed and turn into redundant prediction boxes. To address this issue, our proposed IoU regularization term encourages similar queries to be distinct during the refinement. Through extensive experiments, we demonstrate the effectiveness of our approach in handling challenging scenarios, while incurring only a minor additional computational overhead. The code is publicly available at https://github.com/Eaphan/STEMD.

Multimodal Composition Example Mining for Composed Query Image Retrieval.

Composed query image retrieval task aims to retrieve the target image in the database by a query that composes two different modalities: a reference image and a sentence declaring that some details of the reference image need to be modified and replaced by new elements. Tackling this task needs to learn a multimodal embedding space, which can make semantically similar targets and queries close but dissimilar targets and queries as far away as possible. Most of the existing methods start from the perspective of model structure and design some clever interactive modules to promote the better fusion and embedding of different modalities. However, their learning objectives use conventional query-level examples as negatives while neglecting the composed query's multimodal characteristics, leading to the inadequate utilization of the training data and suboptimal construction of metric space. To this end, in this paper, we propose to improve the learning objective by constructing and mining hard negative examples from the perspective of multimodal fusion. Specifically, we compose the reference image and its logically unpaired sentences rather than paired ones to create component-level negative examples to better use data and enhance the optimization of metric space. In addition, we further propose a new sentence augmentation method to generate more indistinguishable multimodal negative examples from the element level and help the model learn a better metric space. Massive comparison experiments on four real-world datasets confirm the effectiveness of the proposed method.

WeightedSim: privacy-preserving weighted similarity query over encrypted healthcare data

WeightedSim: privacy-preserving weighted similarity query over encrypted healthcare data