Introductory concepts（入门概念）¶
As you get started in the Palantir platform, it can be helpful to think about data in the platform living in two places: the data layer and the object layer.
Data Layer¶
In the data layer, data is stored inside datasets, which typically represent tabular data like you might find in a spreadsheet, but support data at any scale. Datasets usually come from organizational data sources that are synced into the platform, but you can also create your own datasets by uploading approved or notional data.
Every dataset in the Palantir platform maintains a record of how it was produced, so that the origins of data are always preserved and accessible. This concept is known as data lineage.
- Palantir keeps track of which input datasets were used to produce which output datasets. This allows you to always know where a piece of data came from, and understand how data is used.
- Palantir tracks the logic that was applied to produce each output dataset. For example, an input dataset might be filtered to produce a smaller output dataset; that filtering logic is preserved and visible in the platform. There are many ways to write logic in the platform, ranging from code repositories to point-and-click tools.
You can interact with data using one of Palantir's many applications. When you use applications, anything you produce, whether it is a dataset, code, or analysis, is stored in the platform as a resource. Resources are organized into Projects, which serve as permission boundaries for grouping and organizing related work. We will cover the details of how to access and use Projects in one of the upcoming sections.
Object Layer (Ontology)¶
In the object layer, or Ontology, data is stored in objects and links. Objects represent real-world concepts like an airplane, vehicle, or customer, while links represent the relationships between objects. The object layer takes the data stored in tabular datasets—rows and columns of data—and converts it into a series of concepts that anyone in the organization can understand.
In addition to helping make data more understandable, converting data from datasets into objects and links unlocks a broad set of tools for interacting with objects. You can define actions that describe how objects can be changed by people in your organization. This enables you to build applications that access data from objects and capture user decisions back into the system.
The definitions of objects, links, and actions together make up what is called the Ontology, a digital representation of your organization. Developing and using the Ontology to translate data into operational outcomes is a key part of getting value out of the Palantir platform.
中文翻译¶
入门概念¶
在开始使用 Palantir 平台时，将平台中的数据理解为存在于两个位置会很有帮助：数据层(Data Layer) 和 对象层(Object Layer)。
数据层(Data Layer)¶
在数据层中，数据存储在数据集(Datasets)中。数据集通常代表表格数据，类似于电子表格中的内容，但能够支持任意规模的数据。数据集通常来自同步到平台的组织数据源，但您也可以通过上传已批准或模拟数据来创建自己的数据集。
Palantir 平台中的每个数据集都会保留其生成方式的记录，以便数据的来源始终得以保存并可追溯。这一概念称为数据谱系(Data Lineage)。
- Palantir 会追踪哪些输入数据集(Input Datasets)被用于生成哪些输出数据集(Output Datasets)。这使您能够始终了解数据的来源，并理解数据的使用方式。
- Palantir 会记录生成每个输出数据集所应用的逻辑(Logic)。例如，一个输入数据集可能经过筛选(Filtered)以生成一个较小的输出数据集；该筛选逻辑会被保留并在平台中可见。在平台中，有多种编写逻辑的方式，从代码仓库到点击式工具，不一而足。
您可以通过 Palantir 的众多应用程序(Applications)之一与数据进行交互。使用应用程序时，您生成的任何内容（无论是数据集、代码还是分析结果）都会作为资源(Resource)存储在平台中。资源被组织到项目(Projects)中，项目作为权限边界，用于对相关工作进行分析和整理。我们将在后续章节中详细介绍如何访问和使用项目。
对象层(Object Layer)（本体论）¶
在对象层（即本体论(Ontology)）中，数据以对象(Objects)和链接(Links)的形式存储。对象代表现实世界中的概念，例如飞机、车辆或客户，而链接则代表对象之间的关系。对象层将存储在表格数据集中的数据（即行和列）转换为组织中任何人都能理解的一系列概念。
除了帮助使数据更易于理解之外，将数据从数据集转换为对象和链接，还能解锁一系列用于与对象交互的工具。您可以定义操作(Actions)，描述组织中的用户如何更改对象。这使您能够构建应用程序(Applications)，从对象中访问数据，并将用户决策捕获回系统中。
对象、链接和操作的定义共同构成了本体论，即您组织的数字表示。开发和利用本体论将数据转化为运营成果，是从 Palantir 平台中获取价值的关键部分。