Preferences

After uploading your dataset to the Relevance AI's platform, you can check your data, and data statistics and configure some settings.
In this page, we will explain the Preferences feature. The setting page allows you to configure field types, aliases, and more technical settings. To start click on your desired dataset under your account and then "Preferences" from the left side menu.

You will see a page containing the items shown in the following two images.

There are 5 main sections to this page:

Dataset fields

In this section, you can

• Hide one/some fields

If you wish to prevent a field from showing up on dashboards, simply tick Hide for that field

• Assign an alias to a field

Aliases can be used to refer to fields with another name. This feature is very useful in situations like when existing names are too long, not descriptive enough or a simple change of format as shown in the image (i.e. full_name -> Full Name)

• Enter a tooltip for a field

Sometimes not all users are aware of what data each field represent. You can add a tooltip for each field which will show up any time users hover over the field name on a dashboard

• Change the data type if necessary

Data types are detected automatically in our platform and it is best not to change them. However, in the very rare case of wrong type detection, use the marked down-arrow and select the correct data type from the list.

Name of entities in the dataset

The use of the right words makes the insight and the final presentation more tangible and meaningful. If your dataset is on customer reviews, comments, data descriptions or project learnings, you can simply use the right referral to the entities here, (e.i. Reviews, Comments, Descriptions and Learnings respectively). Using the plural word form is recommended.

Main date field

You are provided with date filtering and time series for insight analysis and presentation. Every data entry automatically receives an insert_date_ field, the date the entity has been uploaded to Relevance AI's platform. In case, you wish to analyse with another date field in your dataset, select it from the drop-down menu as shown below.
Note: Keep in mind that date entries must follow dash-separated year, month and day format in 4, 2, and 2 digits (e.g. yyyy-mm-dd). Therefore, in case your dataset contains a date field with values not in the standard format, you need to standardize the date values before uploading your dataset to the platform. For instance,
May 29, 20 --> 2020-05-29
2017-08-21 15:56:48 ---> 2017-08-21

Text highlighting rules

A very useful tool when dealing with text data is a highlighter. Relevance AI's highlighting tool receives a "Text field to highlight" and a "Substring" field where the substring is a subset of the main text. The text subset is highlighted when presenting/visualizing the parent field.
Text field to highlight and the substring field are selected via the marked drop-down menus in the picture below. Leave the weight as none or use the numeric values in your dataset as a weighing parameter if applicable (e.g. when analysing negativity or positivity of comments and darker highlights identify stronger views).

It is also possible to color-code substrings according to sentiment analysis results. More explanation is provided in the next section.

Sentiment range rules

After running sentiment analysis workflow, a few fields will be added to your dataset. Select the sentiment substring and the sentiment values as substring field and weight under Text highlighting rules. Modify the range rules under Sentiment range rules.

Max negative value: any scores smaller than the set value will result in red highlighting

Min positive value: any scores larger than the set value will result in green highlighting