Data tables and time series

Not all knowledge is text. Alongside documents, a knowledge base can hold data tables for structured records and time series for metrics over time. Both live in the knowledge base’s Tables and Time Series tabs.

Data tables

A data table is like a spreadsheet tab with a schema: named columns with a type, optional descriptions, and relations to other tables. Use it for records you want to filter, count and update, such as leads, products, tickets or contacts.

Assistants with the Datatable tool query tables in natural language (“how many open tickets this week?”) and can create or update records. Report templates and scripts in your apps can read them too.

Create a table

1. In the knowledge base, open Tables and click Create Table.

2. Give the table a name and a description of what it contains.

3. Add columns. For each one, set a name, a type and an optional description. Columns can be required, have a default value, be an auto-generated UUID, or be a select with a list of options.

4. Click Create Table.

You can also import a file into a new table: Nebulas infers the columns from the file, and you adjust them before importing.

Work with a table

• Add, edit and delete rows directly in the table.
• Edit schema to add or change columns, in a form or as raw JSON.
• Export CSV to get the data out.
• Index for search to also make the rows searchable by meaning, like documents. Without it, a table is queried, not semantically searched.

Time series

A time series records metrics over time: page views per day, sales per week, response times per service. Use it to track your stats and see how they evolve.

A time series has:

• Keys that identify what is measured, for example user_id or country. A key can’t be changed after creation.
• Metrics, the values you record, each with a name, an optional unit (such as ms or €) and a description.
• Data points, each with a timestamp, its key values and the metric values.

Create a time series

1. In the knowledge base, open Time Series and click Create Time Series.

2. Name it and describe what it tracks.

3. Add its keys and its metrics.

4. Click Create.

Add data points by hand with New row, import them from a file, or have a script or a scheduled task write them. View graph plots the series over a period you choose, and Export CSV downloads the data.

Tip

Time series and data tables are the data sources of report templates. A weekly report can chart a time series and summarize a table, then be sent by email on a schedule. See Create a custom app.