On this page, we will explain how to build up your desired conditions and modifications. Don't forget to review Data type and conditions and Edit types explained at the bottom of this page. To start, select a dataset and go to Editor. Then click on "Rule-based editing" from the menu on the left.
No matter what condition you have in mind, you need to follow three steps under Filter for data where section:
- Select the source field
- Select the applicable rule/condition to the data type in the selected field
- Type in what values you have in mind (if applicable)
followed with AND/OR, repeat the above when using compound or mixed conditionals.
You can see the matching documents immediately appear on the page. For instance, in the example below, we looked for documents whose "Content" field is left untagged and the text under the "Content" field is similar in meaning to "Useless". We intend to introduce a new tag "Useless app" to label these documents in the next step.
Note that you can select all documents or a subset of them using checkboxes on the left.
Under "Set up bulk edits" click on "Add bulk edit". A menu will open on which you can select
- the field to bulk edit
- type of edit (applicable to the selected field)
The image below shows a bulk update where a new tag is to be added to selected entries. Simply click on the Update matching or Update the selected matching documents to apply the changes. You will receive a warning that the changes are not easily reversible.
Note1: Using "And", it is possible to apply more than one update to selected documents
Note2: Check the selected documents under "Selected" on top
Note3: Check all applied changes under "History" on top
Note4: Check recommendations - on top right - which suggests you more edit options based on your selected conditions and documents.
Note5: Adding new tag(s) to a [No Tag] value will automatically remove [No Tag]
Note6: Use "clear all" in Search or Edit section to start over.
Note7: Clicking on values shown on the left menu will add/remove them from the condition/rule under search.
After setting up the filters, scroll down to the table result. Keep in mind AI-tagging results must be included in your filters to be shown in this table. Find the entry whose tagging list you wish to modify and double-click on one tag or next to the last entry. This will activate the editor mode as shown in the image below. You can
- remove any of the existing tags by clicking on the
xnext to them
- select another tag from the dropdown
- type in a new tag and add it to the code frame
Important: Changes will not persist unless you hit the save button on the bottom right.
Values in a dataset are of various types such as strings (e.g. name and text responses), numbers (e.g. prices), and Tag analysis results ( a list of string values). Automatically, rules and conditions applied to different data types become different from one another. For example, when dealing with numbers, we look for values that are greater than another number. Whereas, when working with response sentences greater or smaller that does not make sense. We might want to find sentences containing a specific word or be similar in meaning to another phrase.
Parameters for defining conditions are self-explanatory. The following table summarizes the available condition parameters.
|contains||entries in a dataset whose selected field contain a substring||String||"Conditional Edit" contains "Edit"|
|fuzzy contains||entries in a dataset whose selected field contain parts of a substring||String||"Conditional Edit" contains "Edit"|
|doesn't contain||entries in a dataset whose selected field does not contain a substring||String||"Conditional Edit" contains "AI"|
|is||entries in a dataset whose selected field is equal to ...||All types|
|isn't||entries in a dataset whose selected field is not equal to ...||All types|
|has any value||entries in a dataset that contain a specific field||All types|
|is empty||entries in a dataset that contain a specific field but the field does not have any value||All types|
|is similar in meaning to||only available for vectorized fields and looks for contextual similarity||String|
|contains one of||entries in a dataset that contain at least one of ...||List||"App" exists in [App, UI, useful]|
|doesn't contain any of||entries in a dataset that contain none of ...||List|
|doesn't just contain any of||entries in a dataset that does not just contain a value||List|
|is less than||entries in a dataset whose values under a selected field is smaller than ...||Numeric|
|is greater than||entries in a dataset whose values under a selected field is greater than ...||Numeric|
It is possible to combine conditions and rules with two main parameters, [AND, OR].
Example1: X greater than 5 AND X smaller than 10
Example2: X contains the word "customer service" OR X is similar in meaning to "response"
- AND: for the whole mixed conditional/rule to be valid, every single condition must match. So, X = 7 sets the first example above to True but not X = 11.
- OR: for the whole mixed conditional/rule to be valid, one value single condition is enough. So, X = Very quick reply will activate the condition in Example2. Since even though X does not contain "customer service", X = Very quick reply is similar in meaning to "response".
Depending on the type of values in the selected field for bulk edit, different modification/updates are available on the Bulk Editor. Updates are self explanatory. The following table summarizes the available updates.
|change value to||changes an existing value to another value in matching entries|
|add tags||adds a new tag to matching entries|
|remove tags||removes an existing tag from matching entries|
|merge tags||combines two or more existing tags under a new tag or another existing tag in matching entries|
|rename tags||renames one or more existing tags to a new tag in matching entries|
|update sentiment to||changes sentiment labels in matching entries|
Updated 4 months ago