Daylight | Cloud | April 9, 2022
Enhancement
Active Concepts: Details are displayed for edited active concepts
Science: Properly display concepts written as a possessive when enclosed in quotes
API
The API no longer returns 500 errors when a parameter has too many values
Relevance now always included in concept objects returned by APIs
Active Concepts: Details are displayed for edited active concepts
When a concept has been edited, the details of how the concept has been edited is now displayed in the Active Concepts tab. For example, if the concept beer has been edited to also include brew and ale, then if beer is an active concept, it will display as beer (beer, brew, ale) in the Active Concepts tab.
Science: Properly display concepts written as a possessive when enclosed in quotes
The way that we define the surface form (the way that a concept is displayed in the UI) has been updated for the case where a possessive is enclosed in quotation marks. For example, if “Alzheimer’s” was present in the text, we previously displayed Alzheimer in the UI. This has been updated so that we correctly use Alzheimer’s as the surface form.
The API no longer returns 500 errors when a parameter has too many values
Previously, if an API call was made with an exceedingly large number of values for an input parameter, for example the “filter” parameter of the GET /api/v5/project/<project_id>/docs endpoint, the API call failed and a 500 error was returned. With this release, our API is now able to handle this scenario and successfully return valid responses.
Relevance now always included in concept objects returned by the API
Previously, the relevance score was included in the concept object returned by the API only when the concept_selector parameter was set to “top”. With this release, the relevance score will always be included in the concept object regardless of the input parameters passed to the API call.
Daylight | Cloud | May 7, 2022
Enhancement
Science: Updated clustering algorithm
Science: Updated clustering algorithm
The clustering algorithm is used to identify 7 clusters of concepts within your data and to select 4 representative concepts within each cluster. This information is displayed on the Highlights page and is what you get when you select the “View suggested concepts” option from the Active Concepts tab. The primary change is that the algorithm will now ignore any concept that only appears once within the project.
Daylight | Cloud | April 21, 2022
Enhancement
New Feature: Project Management Page
New Feature: Science Assertions - Ignore Concepts
New Feature: Change Language
New Feature: Project Management Page
We have aggregated all of the information about the project and any action you may want to take on a project in one place. The new project management page can be accessed with the new icon on the bottom left of the application. Click here to learn more about the project management page.
New Feature: Science Assertions - Ignore Concepts
One key feature of the new project management page is the science assertion section. Here, you can tweak the language model that Daylight’s QuickLearn AI engine generates from your data. While Daylight’s ability to quickly and automatically generate the language model specific to your data is a unique value that it delivers, we believe that it’s important to allow subject matter experts, ie. you, to impart your knowledge in order to fine tune the model. To start, we are introducing the ability to Ignore specific concepts which will result in those concepts being removed from the language model. For your desired changes to take effect, you will need to rebuild the project from the footer section on the page to generate a new language model. Click here to learn more about this new powerful feature. Using this feature will have an impact on the analysis results, so we encourage you to discuss your use case with your assigned Customer Success Manager first.
New Feature: Change Language
Another thing that you will be able to do from the Project Management page is to change the language of your project. When building the project, if you select the wrong language for the data that is being analyzed, you no longer have to re-upload the data. You can simply come to the Project Management page and select the correct language and rebuild the project.
How helpful was this article?
Thank you for your feedback!