VectorDatabaseObject

VectorDatabaseObject[…]

表示一个由 CreateVectorDatabase 创建的向量数据库.

VectorDatabaseObject["name"]

represents the database with the specified name in the VectorDatabaseObjects[] list.

VectorDatabaseObject[source]

attempts to recreate a VectorDatabaseObject from source.

更多信息和选项

  • VectorDatabaseObject 表示一个向量数据库实例，允许与存储的向量进行交互和管理.
  • 该符号对象由 CreateVectorDatabase 返回，可实现常见的数据库管理操作，例如添加数据、搜索、列出内容和属性.
  • source 的可能值包括：
  • "name"a named database from VectorDatabaseObjects[]
    File[…]本地文件
    LocalObject[…]本地对象
    CloudObject[…]云对象
  • 可通过 VectorDatabaseObject[…][prop] 获取数据库的属性.
  • 支持以下属性 prop：
  • "Dimensions"存储的数组的数量和数组的大小
    "DistanceFunction"使用的距离函数
    "ID"数据库的 ID
    "Location"数据库被存储在何处
    "Metadata"与每个向量关联的元数据
    "Vectors"存储在数据库中的向量
    "WorkingPrecision"存储和计算的数值精度
    {prop1,…}属性列表
  • When recreating a database, the following options can be specified:
  • GeneratedAssetLocation $GeneratedAssetLocationwhere to save the database
    OverwriteTarget Automaticwhether to overwrite an existing location

范例

打开所有单元关闭所有单元

基本范例  (2)

初始化一个新的向量数据库：

获取数据库的一些属性：

用向量列表初始化一个新的向量数据库：

执行搜索操作：

范围  (6)

Retrieval  (1)

在文件中存储向量数据库：

从文件中重新创建数据库对象：

Recreate the database object from the database name:

Properties  (2)

提取可用属性的列表：

提取存储的向量：

一次提取多个属性：

Metadata  (3)

Create a vector database where each vector is associated with metadata:

Extract vector database metadata:

Extract vector database metadata and other properties:

Create a vector database where each vector is associated with metadata associations:

Extract vector database object metadata:

Extract a specific metadata element:

Extract multiple metadata elements:

Extract metadata and other properties:

Use the object to perform a search operation:

选项  (2)

GeneratedAssetLocation  (1)

创建向量数据库：

将数据库复制到新位置：

删除原始数据库：

OverwriteTarget  (1)

Create a new vector database:

Create a file:

By default, copying the database to an existing file will fail:

Use OverwriteTarget -> True to override the existing file:

属性和关系  (4)

大多数情况下，只能通过名称来引用现有的向量数据库：

相当于：

Create a database with vectors and metadata:

Both vectors and metadata are stored positionally:

A SemanticSearchIndex is backed by a VectorDatabaseObject:

Retrieve the underlying database:

Detailed information on the database can be extracted using vector database properties:

Modifying or deleting the underlying database may cause issues with the index:

Create a file-backed vector database:

Database information is stored in the "WXF" format:

More files could be generated for efficiently storing the vector information:

可能存在的问题  (2)

在指定位置创建数据库：

裸字符串被解释为数据库名称：

使用 File 包装器指定源为文件路径：

Create a path:

Create a database in the specified location:

Move the underlying files:

The original database is no longer usable:

Use VectorDatabaseObject on the moved file with OverwriteTarget True to recreate the database and update the location data:

Wolfram Research (2024)，VectorDatabaseObject，Wolfram 语言函数，https://reference.wolfram.com/language/ref/VectorDatabaseObject.html （更新于 2025 年）.

文本

Wolfram Research (2024)，VectorDatabaseObject，Wolfram 语言函数，https://reference.wolfram.com/language/ref/VectorDatabaseObject.html （更新于 2025 年）.

CMS

Wolfram 语言. 2024. "VectorDatabaseObject." Wolfram 语言与系统参考资料中心. Wolfram Research. 最新版本 2025. https://reference.wolfram.com/language/ref/VectorDatabaseObject.html.

APA

Wolfram 语言. (2024). VectorDatabaseObject. Wolfram 语言与系统参考资料中心. 追溯自 https://reference.wolfram.com/language/ref/VectorDatabaseObject.html 年

BibTeX

@misc{reference.wolfram_2025_vectordatabaseobject, author="Wolfram Research", title="{VectorDatabaseObject}", year="2025", howpublished="\url{https://reference.wolfram.com/language/ref/VectorDatabaseObject.html}", note=[Accessed: 29-May-2025 ]}

BibLaTeX

@online{reference.wolfram_2025_vectordatabaseobject, organization={Wolfram Research}, title={VectorDatabaseObject}, year={2025}, url={https://reference.wolfram.com/language/ref/VectorDatabaseObject.html}, note=[Accessed: 29-May-2025 ]}