Use this guide to kickstart your Daylight experience.
Luminoso Daylight is a text analytics application powered by QuickLearn®, Luminoso’s proprietary natural language modeling system. You don’t need custom ontology libraries or data science experience to use Daylight. As soon as you create a project, Daylight immediately begins analyzing data. Use Daylight to enhance the value of your unstructured natural language text through intelligent, automated analysis and quantification.
Extract actionable insights from what people are saying with Daylight. Organizations use Daylight to analyze text such as open-ended survey responses, product reviews, and support tickets. Using Daylight, they rapidly identify digital problems, understand concepts driving scores, and know which aspects of their products and services hold the strongest sentiment.
Use this guide to familiarize yourself with the basic functionality you’ll need to begin analyzing your data. Once you’re done reading, check out some of the other resources in the Luminoso Help Center.
Luminoso staff create new workspaces and users in Daylight. You will receive an email from email@example.com after your Luminoso Daylight account and user profile are created with an invitation to your new workspace. The link in this email expires after seven days. If your link expires or you can’t find this email after checking your spam folder, contact us at firstname.lastname@example.org.
Click the link to access Daylight. The Daylight login page appears. Note: Use Google Chrome to access Daylight. Microsoft Edge is not a Daylight-compatible browser.
Enter your username, which is automatically set as your email address.
Create a password and enter it a second time for validation.
Bookmark or save the login link for easy access.
Preparing a dataset for upload
Now that you have an account on Daylight and access to a Workspace, the next step is to upload some text data to analyze. You can refer to the Using different data sources with Daylight page for some pointers on different types of text data to consider. You will also want to include some metadata to provide context for the text data, such as demographic information, dates, scores, etc. Metadata can be used to filter the data to analyze a subset of the data set, and numerical data can be used in Drivers analysis, which you will learn about later in this document.
Before we get started, here are some terms that we will be using throughout this section of the document.
A verbatim is the conversational text component of the sample you have collected.
A document is a row of your source data, including the conversational text and any associated metadata.
Metadata is structured data that creates context for text responses. Metadata may include demographics, dates, scores, or product details.
A CSV file, or comma-separated value file, is a plain-text file format that is used to organize data. CSV files exclude styling information that is included in an Excel XLS or XLSX file formats. You can export a CSV file from most spreadsheet editors.
The following table describes the types of data that can be included in the data to be uploaded. The first row of the data will be used as the name of the data in that column. You will need to designate the data type for each column of the uploaded data during the uploading process. There are two ways to do this:
In your dataset file, make sure each header has the data type, then an underscore, then the column name. E.g. score_Rating, string_Location
As you upload your dataset file, you can change individual data types in Daylight.
Column header: text or text_[FieldName]
The natural language verbatims for Daylight to analyze
There must be one and only one Text column per file
Each piece of Text may not exceed 500,000 characters in length
I loved the free coffee and the room was very clean, but it smelled strongly of cigarette smoke.
I booked this room last-minute when my travel plans changed. The price was ok considering it was last-minute but it was way out of the way.
We come to this hotel every year, and we appreciate the consistently top-notch experience!
Column header: title or title_[FieldName]
A way to refer to each document
The Title isn’t analyzed as part of the verbatim, but can help organize documents
Only one Title column is permitted per data file
Titles are optional
We’ll definitely be back!
Column header: string_[FieldName]
Words that help categorize the documents
Can only filter fields in Daylight with up to 10,000 values
Daylight assumes that all dates are in a UTC timezone unless you include an ISO 8601 date with a specific timezone
Helps you filter your project, especially if you upload data more than once
Include as many date columns as needed
ISO 8601 formatted dates:
4/10/18 1:45 PM
Sometimes, a metadata field can have multiple values within a single document. For example, a survey may ask the respondent “which of these products have you tried?”. In such a case, the respondent may select more than one product. There are two ways that you can format the data in such cases:
Have one column for this metadata field and enter all of the values separated with the “|” (pipe) character. In the example above, the column header can be “Products Used” and the value in a given cell could be “ProductA | ProductB | ProductC”.
Have multiple columns with the same column header name with a single value in each cell. In the example above, you would have as many columns as you need with the header name “Products Used” and populate each cell with a single Product. Some of the cells can be left blank.
In both cases, a single metadata field will be created with multiple values.
Supported languages and multilingual datasets
Daylight is capable of performing analysis natively in15 languages. For best results with a multilingual dataset, split your data into one language per upload file. Each language will be uploaded and analyzed as its own Project.
Save as a CSV file
To upload your data to Daylight, save the file in CSV format, and make sure that the file extension is .csv. Daylight will also accept similarly formatted files, such as a tab separated value (TSV) file. In this case, make sure that the file extension is .tsv.
Uploading a dataset to create a new project
To upload a file and create a new Project, go to the Project page and click on the New project button on the top right corner of the page. You can get to the Project page by clicking the Luminoso logo or “Luminoso Daylight” on the top left of your screen. From this page, you can select a file to upload by clicking Choose file or by dragging and dropping a file onto the main panel on the page. Note that you can only upload 1 file at a time. Once a file has been selected, you will see a preview of your data which will look similar to the image below.
Document Usage Meter
The document usage indicator at the top of the page shows document usage relative to the contracted document allotment for your organization. Prior to selecting a file to upload, you will see the licensed document amount and the currently used document count. Once a file has been selected, you will additionally see the number of documents that the selected file will use out of the allotment.
Project Info section
The Project Info section consists of the Project name, Description, Workspace and Language.
Project name - defaults to the name of the uploaded file and can be edited
Description (optional) - can be used for information about the project
Workspace - defaults to whatever Workspace you were using when you clicked New project. If you have multiple Workspaces and prefer a different one, you can change it here.
Language - there is no default language selected. You must select one of the 15 available languages.
Data Summary panel
The Data Summary panel along the left side of the page is where you can designate the data type for each column of your data. There will be an element in this area for each unique column header name. If there are multiple columns with the same column header name, there will only be one element here corresponding to that group of columns.
Use the drop down at the top right of each element to designate the data type. A default value will be set as follows:
If the column header is formatted as described in the Data Fields section, then that information will be used to set the default. For example, if one of the column headers is named “text”, the default for that column will be set to Text. Or if the column header says “date_ResponseDate”, the default for that column will be set to Date.
Otherwise, the data type will default to String.
Each element will also provide information about the values found in that column in the context of the designated data type. For metadata columns, a preview of how this data will appear in the Filters after upload will be shown.
Clicking on an element here will scroll the Data Display panel to bring the corresponding column into view and highlight the column.
Data Display panel
The Data Display panel shows up to the first 100 rows of the data. This can be used as a reminder of the contents of a column as you’re deciding on the data type designations.
Clicking on a column in this area will scroll the Data Summary panel and highlight the corresponding element.
Editing Column Header names
As described in the previous section, the first row of the uploaded data will be used as the name associated with data in each column. You can change the name of any column by clicking on the edit icon next to the column name in either the Data Summary panel or the Data Display panel.
When multiple columns have the same name, they will be combined into a single data field. You can edit column names to combine columns or separate combined columns.
Errors and Warnings
At the top of the Data Summary panel, a count of Errors and Warnings found in the data will be shown. Errors must be corrected before the project can be created. Warnings are things you should take a look at, but will not prevent you from creating the project. The Create project button will not be enabled until all Errors have been corrected. The counts shown will change as Errors and Warnings are corrected or new ones introduced.
Errors and Warnings that apply to the entire project will be shown just below the Errors and Warnings counts. They will be automatically dismissed as the issues are resolved.
Errors and Warnings specific to a column of data will be shown in the corresponding element in the Data Summary panel.
Note: some Errors and Warnings cannot be resolved within Daylight. In this case, you will have to make edits to your CSV file and upload the newer version.
Complete the upload
Once you have filled in the Project Info section, confirmed the data type designations, corrected all Errors and reviewed the Warnings, you can complete the upload process by clicking the Create project button on the bottom left. Just above the button, you will see a checkbox if you would like to receive an email when the analysis has been completed and the project is ready to be viewed. It is checked by default.
Uploading to an existing project
The steps described in this section can be used to upload additional data to an existing project as well. To do this, click Upload data from either the Project’s Highlights page or within the Project tile on the Projects page. When there are mismatches between the data field names in the project and the column headers for the data being uploaded, Warnings will be shown to highlight these mismatches.
Start with the Highlights feature
The Highlights feature is a starting point in understanding the contents of your project. In this feature, orient yourself and start navigating the primary questions you can answer using Daylight:
Theme detection — What are people talking about?
Emotional analysis — What are the overall and specific sentiments?
Differential analysis — What differences can Daylight identify in your metadata?
Score drivers — What is driving your customer experience up or down?
You’ll first view Daylight’s answers to these questions in Highlights, and then engage deeply with each data perspective as you move through the Volume, Sentiment, Drivers, and Galaxy features.
Display Highlights: In the Projects list, click on the name of the project you want to view. The Highlights feature appears.
Make immediate observations about your data in the Insights from your project section:
Which concepts are most prevalent? — View which concept groups appear most frequently in your dataset, as compared to the native language as a whole. For a deeper look at prevalence, click Volume feature to access the Volume view or select the Volume icon in the sidebar.
What do people feel strongly about? — View a preview of some of the concepts in your dataset that provoked a strong reaction. For a deeper look at sentiment, click Sentiment feature or select the Sentiment icon in the sidebar.
What issues are affecting my scores? — View how certain concepts affect scores for your project. If you did not include a column titled score in your metadata, this section is blank. If you have multiple saved scores, you can toggle between by selecting a field with score data in the upper right of the card. For a deeper look at your score drivers, click Drivers feature or select the Drivers icon in the sidebar.
What are the largest clusters of conversation? — View the largest concept groups that Daylight identified in your project. Concept groups are colored to help you identify them at a glance. For a deeper look at your concepts, click Galaxy feature or select the Galaxy icon in the sidebar.
Use the upper section of the page to manage your project:
Edit name & description — Click and type here to update your project’s title and description.
Project details — View who created your project and when they created it. You can share or delete your project in this section.
How many documents are in this project? — View how many documents are in your project. You can upload more documents or make a copy of your project here.
What metadata do my documents have? —View the amount and type of metadata your project includes.
Learn the sidebar
Each feature, except for Highlights, shares a common sidebar that you can manage and configure your Daylight information. Use it to:
View search details.
Edit your active concept list.
Edit and manage your shared concept lists.
Apply metadata filters to your project.
Configure each feature visualization. Note: This option is not available in the Galaxy.
View all documents that match your search and filter selections.
Download analysis information to a spreadsheet or copy a direct URL to your clipboard.
To see information in the Search details tab, either enter a search in the search bar or click a concept to select it. Then, the tab displays total, exact, and conceptual match information for the active search.
If you use a minus sign to remove a concept from your search results, like “coffee, -tea,” you also see an Excluded matches section.
Use the Active concepts tab to view and manage suggested or active concepts and to create and edit shared concept lists. If you’re viewing a new project, click Add to active concepts to begin your analysis using concepts that Luminoso’s AI suggested, or Dismiss to clear. To begin using only some of your suggested concepts, clear the check boxes to dismiss the concepts you don’t want to keep and click Keep selected when you’re done.
The Active concepts tab view after loading a shared concept list into the active concept list.
You can view all suggested concepts again by clicking the three dots at the top of the Active concepts pane. Daylight identifies these concepts as potentially relevant based on their prevalence and relationship to other concepts in your project.
As you add more concepts to your active concept list, they appear in this tab. To edit a saved concept, select it and click the pencil icon in the search bar.
Use the Filter documents tab to select filters that change what data displays in the feature view. When you select a filter, it remains consistent across features. This is also true if you share a direct link to the page you’re viewing with another Daylight user.
View the number of total documents in your filter in the Current filter section. Use the reset button to clear all filters you’ve applied.
Use the Configure visualization tab to select concepts to view in visualizations, count matching documents, and select feature configuration options. This tab is not available in the Galaxy. The options you find in this tab vary slightly by feature.
The Configure visualization options in Volume
Additionally, you can access the Unique to this filter concept selection feature in the Configure visualization tab. Use Unique to this filter to isolate segments within projects and find concepts that are unique to that segment, relative to the rest of the project. For example, if a project includes the entire U.S., but you filter for only customers in Utah and apply Unique to this filter, you view the concepts that are unique to Utah customers compared to the other 49 states.
In the Documents tab, view documents that match any concept and filter selections you have applied. If you select a concept, the tab displays only documents that contain exact and conceptual matches for that concept. Use the exact or conceptual matches icons to view only exact or conceptual matches icons in the Showing section. When you apply a filter, the tab displays only documents that match those filter criteria.
For each document that meet the filter and concept selection criteria, the Title of the document, a list of selected metadata associated with the document and the document itself are displayed.
Document viewer modal
Clicking on a document in the side panel opens that document in a modal. Here, you can see the full text of the document and the associated metadata for the document. Clicking on the up and down arrows along the left will cycle through the documents shown in the side panel. Alternatively, you can scroll through the documents in the side panel and click on a different document to view in the modal.
Selecting metadata fields
You can select which metadata fields to display in the side panel view and the document viewer modal by clicking Select fields. You can make independent selections for the side panel and the viewer modal. The example below will display all of the metadata fields in the viewer modal, but only 3 of the fields in the side panel.
Click Download to export a spreadsheet that contains all documents that match your concept and filter selections. Then, you can read all documents that interest you.
Creating a branched project
Click create project to build a new Daylight project based on your concept and filter selections. This will create a new project that only contains the documents that match your concept and filter selections.
Export & share
Use the Export & share tab to manage your project’s exports.
Click Copy link to copy to copy a direct link to the project feature (Volume, Sentiment, Drivers, or Galaxy) that you are viewing to your clipboard. You can share this link with other Daylight users or use it to access the same view later.
In the Export current view section, click Export to download an Excel spreadsheet that contains the data for the feature and configuration that you’re viewing. This spreadsheet reflects any filter information that you have applied. Note: This option is not available in the Galaxy.
In the Export raw data as spreadsheet section, click Export to download one of the options:
Current documents – Downloads a spreadsheet that contains all documents included in your search and filter configuration. After clicking Export, choose Total matches to download both exact and conceptual matches, or Exact matches only to download only exact matches for your search. Read our extended documentation on the Downloads tab for detailed information on the selections that affect your download.
Concept-Filter association scores – Downloads a spreadsheet that includes the association scores between your active concepts and metadata filters.
Concept-Concept association scores – Downloads a spreadsheet that contains the association scores that each active concept has with every other active concept.
Current View Box
The Current View Box located at the top left corner provides information about what you are currently visualizing.
In this example, the Active Concepts are being used in the visualization and “Example Concept List” is the Shared Concept List which has been loaded. Also, from the Filter tab, the value “3” is selected from the “Overall score” metadata field. This filter setting results in 1,095 (which is 12.5%) documents being represented in the visualization.
As you explore and make discoveries, you can save and share the settings for visualizations. This will allow you to easily save your work and get back to an interesting visualization or share key insights with your colleagues.
To Share a view, click on the “Share” button located within the blue header of the Current View Box to open the popover. Give your View a name and click “Create” to save the View.
This will save all of the settings that make up the visualization including the Feature that you are on (ie. Volume, Sentiment, etc.), the Shared Concept List being used, the Configure Visualization selection and Filter settings.
To load a Shared View, use the drop down list from the Current View Box to select a View to load.
To make any changes to or to delete a View, click “Share” to open the popover, then select a View to edit/delete from the “Edit” drop down. Doing so will open the below popover.
Here, you can
change the name of the View with the pencil icon.
If you have made any changes since loading the View, you can “Update” the View with the current configuration
The itemized list shows the differences between the current state and the loaded View
Unchecking the checkbox above the itemized list will stop showing the differences so that you can see the detailed settings of the currently loaded View
The trash icon in the bottom right will delete the View
Search bar and concept editor
Use the search bar, which sits at the top of all features except Highlights, to search for concepts. In this bar, you can also edit a concept’s name and color and create and edit compound concepts.
Select a concept by entering the concept in the search bar and or clicking to select it in a feature. To edit a concept, save it by clicking + Add to active concepts. Then, click the edit icon to open the Edit active concepts editor. Use the editor to:
Change the concept’s label.
Add additional concepts to create an advanced concept.
Update the color assigned to the concept.
When you save a compound concept, Daylight automatically assigns your search term concepts as a label, which does not update when you add or remove further concepts. We recommend updating the label to something that’s easy to remember and find later.
Volume feature: Quantify concepts
Gain quick insight into concept representation across the dataset. The Volume feature automatically quantifies how frequently concepts are referenced in the project using the exact and conceptual matching identified by QuickLearn.
Display Volume: Click the Volume icon in the sidebar or select Volume feature on the Highlights page. The Volume feature appears. Browse the concepts. If you’re viewing a newly created project, no filters are applied from your project’s metadata in the Filter documentspane and the sorting filters are defaulted to their presets.
Use the arrows at the top of columns to change the order of your concepts and use the Configure volume visualization pane to adjust how concepts are displayed:
Which concepts to visualize — Select Top concepts, Active concepts, or Unique to this filter from the menu. You can also select the number of top concepts you want to view.
How to count matching documents — Select Include conceptual matches if you want to view exact and conceptual matches. Leave the box unselected if you want to view exact matches only.
How to visualize concept volume — Select the sort order you want:
Overall concept volume — Sorts by the most uniquely represented concepts in the dataset compared to everyday conversation, raising highly represented concepts and pushing down common terms.
Over time — Sorts based on the first letter in concept.
By category — Sorts by raw frequency of either Total matches or Exact matches that you selected in Select match type.
By score value — Sorts by score values, if available.
Across a numeric range — Sorts by a numeric range, if available.
Click the save concept icon + to save a concept. If you haven’t saved any concepts yet, every concept displays the save concept icon next to it.
Galaxy feature: Visualize associations
Use the Galaxy feature to see how strong conceptual relationships are in the context of your entire project. The size of a concept in the visualization indicates how relevant it is in the project. Related concepts cluster together by theme. Since the view is dynamic, you can drag concepts in the space to realign the axis. Highly related concepts group more closely while less related concepts rotate away.
Display the Galaxy: Click the Galaxy icon in the sidebar or select Galaxy feature on the Highlights page. The Galaxy opens.
Use the Galaxy as a visual overview of your data. Click, hold, and drag a concept to rotate the Galaxy and view the 300 dimensions the Galaxy contains. As soon as you stop holding the concept, the Galaxy collapses and displays as two-dimensional again, although it is now reoriented.
Click the reset button to restore the Galaxy to its default orientation.
Click the X and Y axis icons to orient the Galaxy around specific active concepts. If you haven’t saved any concepts yet, you can’t use this option.
Once either the X or Y axis are set, the grey axis toggle button turns blue. Click the button to turn on the axis label. When selected, the button is green.
Click the Active concepts tab to view specific concepts. If you haven’t saved any concepts yet, this tab displays suggested relevant concepts that Daylight identified when you created your project. Daylight automatically suggests groupings of related concepts and assigns colors to demonstrate association.
Select a concept in the Galaxy or the Active concepts pane to view its concept details. The concept appears in the Concept details pane and displays as orange in the Galaxy, while related concepts turn blue and unrelated concepts turn gray.
Click the X next to the concept’s name to deselect it. The concept disappears from the Concept details pane and Galaxy’s colors return.
Create a new project from selected data
After you spend some time with your data, consider using Daylight’s learning ability to gain new insights from a smaller data set. When you create a new project from filtered data, you include only the data you selected in your filters. QuickLearn then recalculates the association between concepts, giving you new calculations that are displayed in the Volume, Sentiment, Drivers, and Galaxy features. The new project retains the metadata you added when you created the original project.
The Create project from selected matches option is available in every feature.
Filter your dataset to show the subset you want to analyze.
Click Create project. A Create project from matching documents window appears.
Add a name your project if you don’t want to keep the automatic one and optionally add a description.
Click Create project. Select Send me an email when this project is finished building if you’d like a notification when the project is uploaded. The Create project button turns gray and the project begins uploading.
Stay in the window until it changes to notify you that your new project is being built.
Click Back to Projects to view your other projects or click Close to return to your original project.
Access your new project from the Projects page when it’s done uploading.
Drivers feature: Analyze mixed datasets
In the Drivers feature, view how your structured numeric data relates to unstructured natural language concepts and reveals why scores increase or decrease. See how concepts in subsets of your data raise or lower overall satisfaction scores for your product. The graph allows you to identify trends in a glance, and the filtering criteria allow you to laser-focus for deeper insights.
Display the Drivers feature: Click the Drivers icon in the sidebar or select Drivers feature on the Highlights page.
Use the arrows at the top of columns to change the order of your concepts and use options in the Configure visualization tab to adjust how concepts are displayed
Use the Filter documents tab to apply filters to your visualization. The average score line, which runs horizontally across the graph, updates as you change your filter criteria. Note: You can’t use the value you selected as a score as a filter.
Use the cursor to zoom in and out on the concepts you see in the graph.
Click on a concept’s colored dot to see more details about it. The concept opens in the Concept details pane.
To use trend analysis in Drivers
Open Daylight and select the Drivers feature.
In the Configure visualization tab, select a score field from the menu.
Choose Autosuggest, Saved concepts, or Unique to this filter from the Which concepts to visualize section. Concepts that Daylight identified as strongly influencing your score are auto suggested. Alternately, save concepts that are relevant to your business to view them in Drivers. Finally, use the Unique to this filter option to identify auto suggested concepts that are uniquely relevant to any filters you’ve applied, as compared to the rest of your dataset.
Click the pencil icon next to Select date field to compare periods and select your desired date field from the menu. The visualization splits into two scatterplots. The top scatterplot automatically displays the full date range for the filter you applied.
Refine the date range in the top scatterplot to an interval you’re interested in. The bottom scatterplot automatically identifies and displays the same size time interval ending a day before the interval you selected for the top pane. For instance, if you selected a 10-day interval of time, the bottom scatterplot would display the previous 10-day interval.
Change the date interval in the lower scatterplot, if desired. Each scatterplot displays the most relevant auto suggested or saved concepts for that period of time.
To configure Drivers concepts using the concept tab
As part of this change, we also introduce a Concepts tab at the top of the Drivers visualization. Use this tab to select the concepts you view in the single drivers visualization, or independently for the top and bottom comparative visualizations.
Click the Concepts tab. It opens a list of concepts that match whatever score field and type of concepts to visualize you selected, sorted by absolute difference.
Configure the concepts to match what you want to see in the scatter plot visualization:
Select a score field option from the menu. 50 autosuggested drivers automatically appear in the scatterplot.
Limit the number of autosuggested drives you see to 10, 20, or 50, using the menu in the Autosuggest row.
Select the Active concepts option to view concepts you’ve added to your list in the Active concepts tab. Add or remove concepts from this list and see your updates immediately reflected in Drivers.
In Concepts (Below), toggle on Same concepts as top scatterplot to match the concepts that appear in the bottom scatterplot to the concepts that appear in the top scatterplot. If you have selected autosuggest and the AI determines that a concept was not relevant during the previous period or year over year period, it does not appear in the bottom scatterplot.
Click Visualization to reopen the Drivers scatterplot and view your selected concepts.
To organize concepts in separate Drivers trend analysis panes
Make sure the Drivers visualization is showing the dates you want to use, and that you’ve selected either Autosuggest or Saved Concepts in the Configure visualization tab.
Click Concepts (Above) or Concepts (Below) to edit the concepts you view in each scatterplot. Depending on your needs, you may wish to view the same or different concepts in the top and bottom panes. Sentiment feature: Identify positive, negative, and mixed sentiment
Sentiment feature: Detect positive, negative, and neutral
Use Sentiment to learn about concepts’ association with types of emotion. Daylight uses a proprietary deep learning model to determine sentiment for all concepts in a project and assigns a label of positive, negative, or neutral, based on the concept’s context. Please use the percentage counts in this feature as a general indication.
Display Sentiment: Click the Sentiment icon in the sidebar or select Sentiment feature on the Highlights page. The Sentiment feature appears.
Use the arrows at the top of columns to change the order of your concepts and use the Configure visualization tab to adjust how concepts are displayed.
Hover over concept’s negative and positive match breakdown to view an additional percentage of neutral matches.
Click a concept row. The bar expands to display examples of the concept’s use in positive, negative, and neutral contexts, and the concept displays as selected in the Concept details pane. Tip: If the concept is interesting, save it to access it easily for analysis in all features.
Click Download more matches to create an export where you can review all matches for the sentiment concept.
To close the sentiment example bar, deselect the concept in the Concept details pane.