Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors
post
page
Filter by Categories
Data Scraping
Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors
post
page
Filter by Categories
Data Scraping

Project management

This section documents Daylight’s project management page.
Learn about various project settings and how to edit science assertions.

Who can access this page?

Account type: all except trial accounts

ViewerEditorCreatorAdmin
View everythingXXXX
Copy projectXXXX
Change name/descriptionXXX
Rebuild projectXXX
Change languageXXX
Get prompted to rebuildWhen new science and/or languageWhen new science and/or languageX
Delete projectXXX
Rebuild if sentiment now supportedXXX
Upload dataXX
Edit science assertionsXX

The permissions above apply both at the organization and workspace level. Organization-level permissions apply to all projects across all workspaces. Workspace-level permissions only apply to projects within a given workspace.

Warning: Exercise caution when granting org-level permissions. Recipients of such permissions will have visibility into all projects without exceptions. Double-check to ensure you’re not exposing sensitive information to the wrong users.

Next build

Review science assertions for the next build and trigger a rebuild in this section.

This section contains:

  1. Science assertion summary
  2. Rebuild button
  3. Estimated rebuild time

Science assertion summary

Rebuild button

This button triggers a rebuild of your project.

The option to rebuild is only available if there are changes that require rebuilding, such as:

  1. There are pending science assertions
  2. There is a pending science update
  3. There is a pending project language change

Estimated rebuild time

As the name suggests, this is a rough estimation of the time it will take for your project to rebuild.

Language

You have the ability to change the language Daylight uses to analyze your dataset in this section.

Open the dropdown menu and select your desired language. Click on the “update at next build” button to queue it up for the next build. Daylight will then show you a summary of the language change in the tile.

warning: Changing the language of your project alters how Daylight analyzes your data. Do not change the language unless you are sure the current language setting is wrong.

Page overview

Science assertions

The Science Assertions feature allows subject matter experts to assert their knowledge and expertise into the construction of Daylight’s language model. Expert users can make adjustments by instructing QuickLearn, our core AI engine, to ignore specific concepts, for example, which results in those concepts being excluded from the resulting language model.

Using the Science Assertions feature does change the fundamental ways by which a project is analyzed. Sometimes that’s important for specific datasets, but it can also introduce bias if not done carefully.

Ignore

This assertion removes a concept from the available list of concepts for the project. For example the concept “drink” might be unhelpful in a beverage review dataset because all of the data relates to drinks. In this case, it might be a good idea to ignore “drink” so more interesting concepts will be easier to notice. If ‘drink’ is not a concept, the application will continue to ignore that assertion during rebuild.

Edit button

This button opens the assertion edit modal where you can edit science assertions.

edit button

Rebuild button

This button triggers a rebuild of your project.

rebuild button

Assertion edit modal

This assertion removes a concept from the available list of concepts for the project. For example the concept ‘drink’ might be unhelpful in a beverage review dataset because all of the data relates to drinks. In this case, it might be a good idea to ignore ‘drink’ so more interesting concepts will be easier to notice. If ‘drink’ is a concept, Daylight will remove it from the list of concepts during rebuild. If ‘drink’ is not a concept, the application will ignore that assertion during rebuild.

Assertion edit modal

assertion edit modal diagram

This interface enables you to add, delete and edit assertions.

The main section are:

  • 1Current build
  • 2Assertion editor
  • 3Next build
  • 4Save & close

Current build

This is the list of science assertions used to build or rebuild your project with.

current build list

Next build

This is the list of assertions queued up for the next build. They will be in effect after your project rebuilds.

next build list

Assertion editor

assertion editor diagram

This is where you input new assertions and delete recently added assertions.

It has three main parts:

  • 1Assertions list
  • 2Add button
  • 3Assertion input field

Assertion input field

Type words to ignore in this text field. As you type, Daylight will present you with a list of auto-suggested concepts. Click on an auto-suggested concept to add it to your ignore list. Hitting the enter key or clicking the add button also adds the term you typed to your assertions list.

assertion input field

The auto-suggested concepts are the most prevalent concepts in your project. Your concept does not have to be in the list and you’re welcome to type freely.

adding and editing ignore assertions

Delete a term from the assertions list by clicking on the trash icon next to it.

deleting current ignore assertions

Assertions list

Ignore terms populate here. The most recent goes on top. Concepts have a trash can icon next to them. Clicking on the icon will remove that concept from the list, so it will not be in the next build.

assertions list

Next build

The main parts are:

  • 1Reset button
  • 2Next build list

Reset button

This button removes all the pending assertions for the next build.

reset changes button

Next build list

This section displays the list of pending assertions for the next build. Daylight will execute these assertions during rebuild if they are valid.

next build list

Save and close

As the name suggests, this button saves your new science assertions to your project. It then closes the modal.

save and close button

Valid ignore assertions

An ignore assertion is valid if the term you’re trying to ignore is a concept in your project. A term might not be a concept if it never comes up in your dataset. Or if Daylight considers it to be insignificant to the dataset as a whole. If that is the case, there will be no functional change to the project, though the assertion will be remembered and apply if more data is uploaded which contains the concept.

Related Posts