Natural Language

In this document we will review best practices when setting up Kolena datasets for NLP or LLM problems.

Basics

Supported File Data Formats

The Kolena SDK supports uploading of data in the Pandas DataFrame format.

The Kolena web app supports the following file formats.

Format	Description
.csv	Comma-separated values file, ideal for tabular data.
.parquet	Apache Parquet format, efficient for columnar storage.
.jsonl	JSON Lines format, suitable for handling nested data.

Using the text column

Text samples can be visualized on Kolena one of two ways.

Gallery mode: visualizes each text value as a tile.

To enable this view, include your primary text values in your .CSV in a column named text.

Gallery View

Example

The Text Summarization ↗ example showcases how texts can be uploaded in Gallery mode.

Tip

Use the TextSegment or LabeledTextSegment annotations to highlight parts of your text that is of interest to you.

Tabular mode: visualizes each text field with its corresponding meta-data in a table with common table functionalities.

To use this view, simply provide the text values in your .CSV with any column names other than text.

Tabular View

Using fields

You can add additional information about your text by adding columns to the .CSV file with the meta-data name and values in each row.

Tip

Kolena is able to automatically extract multiple properties from your text values by Extracting Metadata from Text Fields. You can use these values to create test cases and better understand your data.

Uploading Model Results

Model results contain your model inferences as well as any custom metrics that you wish to monitor on Kolena. The data structure of model results is very similar to the structure of a dataset.

• make sure to link your inferences to the dataset using the same unique ID (for example the text field) you used when uploading the dataset.
• use ScoredTextSegment or ScoredLabeledTextSegment annotations to indicate the inference confidence score.