Chuan Hu is a PhD student at the Computer Network Information Center of Chinese Academy of Sciences and University of Chinese Academy of Sciences. His research focuses on graph database, specifically in areas such as graph query, query optimization, and multi-modal query. In the field of data system, he works on data integration, multi-store system, and cross-model query processing.
Interactive Graph provides a web-based interactive operating framwork for large graph data, which may come from a GSON file, or an online Neo4j graph database.
PandaDB is a high-performance graph database that provide Cypher support, allowing for efficient querying and manipulation of large and complex graph datasets.
Lynx is a general graph query framework to simplify querying graph data by converting complex statements into basic graph operations.
PiFlow is an easy to use, strong scalability, powerful big data pipeline system developed based on the distributed computing framework Spark.
Lynx is a flexible graph query framework designed to streamline the process of querying graph data across diverse data sources. By abstracting complex query statements into basic graph operations, Lynx allows developers to retrieve data through user-defined interfaces rather than direct connections to data sources. This approach simplifies the integration of heterogeneous data sources, offering a robust and generic foundation for building graph query engines.
The paper addresses the increasing demand for handling hybrid queries in graph databases, especially in applications like social networks and smart cities. PandaDB, the proposed AI-native graph database, offers a unified approach to managing both structured and unstructured data. Key features include the online extraction and indexing of semantic information from unstructured data, as well as optimization strategies for hybrid queries.
This research introduces KVGDB, a lightweight and scalable graph database designed to handle varying scales of graph data, from small to large-scale datasets.
Built on key-value storage and implemented using RocksDB, KVGDB efficiently maps graph data to key-value structures, enabling effective management and query processing across different scales.
The database supports both embedded usage and distribution across environments, making it adaptable for a wide range of applications
This research tackles the challenge of querying multimodal data, which includes both structured data and unstructured forms such as audio, images, and videos.
This research contains the following contributions:
This research presents SciDG, a benchmark framework designed to evaluate the performance of graph database systems managing dynamic graph data, which is critical for scientific applications. SciDG assesses how storage structures influence query latency, particularly for version-related queries, and helps developers choose the most effective storage solutions.
This research introduces BIT, an innovative indexing method aimed at improving the performance of computational tasks in NCBI Taxonomy, widely used in biomedical and ecological research.
BIT encodes tree-like structures into bit-vectors using the Polychotomic encoding algorithm, storing them efficiently in bitmaps. By leveraging parallel bit operations, BIT significantly accelerates tasks such as finding the lowest common ancestor and listing descendants.
Experimental results show that BIT outperforms existing tools, offering a faster solution for managing large-scale taxonomy data.
MMDBench introduces a comprehensive benchmark program designed to assess the performance of hybrid queries in multimodal databases. The benchmark includes a data generator, a query workload, and a unified integration framework. Specifically targeting social network scenarios, the data generator produces multimodal data, while the query workload replicates typical operations that query both structured and unstructured data.
