execute-workflows-1## When to split text
When working with free text, survey comments for instance, we are dealing with text pieces that can be composed of multiple sentences and more importantly sentences that are about totally different topics. The following piece is a response to "What else do you expect from the politicians?"
More focus on fact checking. Increase the budget for public education. Better age-care system.
It is clear that even though it is one entry (i.e. a piece of text written by One person) in the dataset, there are three action points associated with it.
* More focus on fact checking. * Increase the budget for public education. * Better age-care system.
Split text helps with such a scenario and the results allows a more fine-grained analysis of the data.
Relevance AI provides you with a workflow that reads a specified text field from your selected dataset, and breaks the text pieces into their composing sentences. These sentences are then treated as independent entries and saved into a new dataset. Note that each row in the new dataset contains all other fields associated with the source piece of text.
Once you have uploaded your data, steps for splitting text into sentences are:
- When on the main page, go to "Workflows" and locate and click on "Split text into sentences".
- The setup page will open. Click on "Get started". Choose the field you wish to process. Specify a name for the resulting dataset as well as a name for the new field (e.g. sentences) in the new dataset; this is the field (i.e. column name) in which the resulting sentences are saved.
- Finally, execute the workflow using your preferred mode.
You will be presented with a table to view the progress. Once marked as completed, go to the Dataset page and select the new dataset.
Updated about 6 hours ago