Cardinality estimation is one of the most vital components in the query optimizer, which has been extensively studied recently. On one hand, traditional cardinality estimators, such as histograms and sampling methods, struggle to capture the correlations between multiple tables. On the other hand, current learning-based methods still suffer from the feature extraction of complex predicates and join relations, which will lead to inaccurate cost estimation, eventually a sub-optimal execution plan. To address these challenges, we present a novel end-to-end architecture leveraging deep learning to provide high-quality cardinality estimation. We exploit an effective feature extraction technique, which can fully make use of the structure of tables, join conditions and predicates. Besides, we use sampling-based technique to construct sample bitmaps for the tables and join conditions respectively. We also utilize the characteristics of predicate tree combined with recursive neural network to extract deep-level features of complex predicates. Finally, we embed these feature vectors into the model, which consists of three components: a recursive neural network, a graph convolutional neural network (GCN) and a multi-set convolutional neural network, to obtain the estimated cardinality. Extensive results conducted on real-world workloads demonstrate that our approach can achieve significant improvement in accuracy and be extended to queries with complex semantics.

Zhi Wang

Hancong Duan

Yamin Cheng

Geyong Min

Information Systems

信息系统是支持数据密集型应用程序的软件和硬件系统。 《信息系统》杂志发表有关信息系统语言、数据模型、流程模型、算法、软件和硬件的设计和实施的文章。主题领域包括主要国际数据库会议（例如 ACM SIGMOD/PODS）中提出的数据管理问题、VLDB、ICDE 和 ICDT/EDBT）以及数据挖掘/机器学习、与结构化数据协调的信息检索、互联网和云数据管理、业务流程管理、网络语义、视觉和音频信息等领域的数据相关问题系统、科学计算和数据科学。与大规模并行数据管理、实践中的容错以及数据密集型系统的专用硬件有关的实施论文也受到欢迎。也欢迎来自城市信息学、社会和自然科学、物联网等应用领域的手稿。所有论文都应突出数据管理问题的创新解决方案，例如新的数据模型、性能增强，并展示这些创新如何有助于实现应用程序的目标。

Information systems are software and hardware systems that support data-intensive applications. The journal "Lecture Notes in Business Information Processing" publishes articles related to the design and implementation of information system languages, data models, process models, algorithms, software, and hardware. The subject areas include data management issues presented at major international database conferences (such as ACM SIGMOD/PODS), VLDB, ICDE, and ICDT/EDBT, as well as data-related issues in fields like data mining/machine learning, information retrieval coordinated with structured data, internet and cloud data management, business process management, network semantics, visual and audio information systems, scientific computing, and data science. Implementation papers related to large-scale parallel data management, fault tolerance in practice, and dedicated hardware for data-intensive systems are also welcome. Manuscripts from application areas such as urban informatics, social sciences, natural sciences, and the Internet of Things are encouraged. All papers should highlight innovative solutions to data management problems, such as new data models and performance enhancements, and demonstrate how these innovations contribute to achieving the goals of applications.

信息系统是支持数据密集型应用程序的软件和硬件系统。《Lecture Notes in Business Information Processing》期刊发表有关信息系统语言、数据模型、流程模型、算法、软件和硬件设计与实施的文章。主题领域包括主要国际数据库会议（如ACM SIGMOD/PODS）中提出的数据管理问题、VLDB、ICDE和ICDT/EDBT，以及数据挖掘/机器学习、与结构化数据协调的信息检索、互联网和云数据管理、业务流程管理、网络语义、视觉和音频信息等领域的数据相关问题。与大规模并行数据管理、实践中的容错以及数据密集型系统的专用硬件相关的实施论文也受到欢迎。此外，来自城市信息学、社会科学、自然科学和物联网等应用领域的手稿也被鼓励提交。所有论文应突出数据管理问题的创新解决方案，例如新的数据模型和性能增强，并展示这些创新如何有助于实现应用程序的目标。

Learning complex predicates for cardinality estimation using recursive neural networks

ivySCI AI Smartly Parses PDF, Answers Researchers' Questions, and Helps You Understand Papers in Seconds

Journal Info

Category	Quartile
COMPUTER SCIENCE, INFORMATION SYSTEMS	2