Search Improvements: Rankings and Special Characters

Search is a core feature of data.world and is consistently a focus of our improvement efforts. This month, we've rolled out improvements to result rankings on searches that contain more than one word. This improvement provides better rankings for results with titles that exactly match the submitted search as well as several other ranking and relevance improvements. These changes also provide better support for searches that contain special characters such as ampersands, dashes, underscores, and slashes.


March 2021 monthly product update

Here is a roundup of some of the key features we released last month, February 2021 -- Looker metadata support, custom metadata field default edit behavior improvements, Gra.fo enhancements, and dataset sync webhook notifications.

Gra.fo navigation and linking improvements

Check out recent UX improvements to Gra.fo, our visual knowledge graph modeling tool. 

  1. To quickly zoom out to see the full graph, use the keyboard shortcut:
    OPTION (ALT) + SHIFT + 0
  2. To zoom to a specific concept:
    OPTION (ALT) + SHIFT + click concept
  3. You can now deeplink directly to a concept in the graph using the Copy link to concept option in the concept menu. Watch the video to see how we use this feature to move fluidly between a data.world catalog and Gra.fo.

Gra.fo is a companion product to data.world with a 30 day free trial of premium team features. Visit Gra.fo to learn more, or reach out to help@data.world.

February 2021 monthly product update

Here is a roundup of some of the key features we released last month, January 2021 -- broader resource support for recently viewed resources and auto suggestions from the search bar, collection search filter/facet, collection management API, and enhanced SQL support for windowed aggregations.

New Dataset Sync Notifications

data.world now sends 2 additional webhooks: Dataset Sync Failure and Dataset Sync Recovery Success

When a dataset within your org fails to sync, data.world will immediately send a webhook message to the configured URL with details about the failure, including error code and error messages from the downstream source.

With those sync failure details, you could configure a ticket to be created in your ticketing system, trigger an alert for a dev-ops team, or simply just make the team aware of the issue.

Natively, data.world will format the webhook payload to post messages in Slack when it recognizes the configured webhook is Slack, which is shown in the demo video below.

Demo


Screenshots


Technical Details

syncTask.status.success is triggered only after the recovery from a failure. 

Example payload, in JSON:

{
  "datasetid": "sync-demo",
  "event": {
    "type": "SUCCESS",
    "task": "state_table.csv",
    "created": "2021-02-04T06:37:06.936Z"
  },
  "eventType": "syncTask.status.success",
  "resourceUrl": "https://data.world/texas-whiskey/sync-demo/file/state_table.csv",
  "resourceType": "URL",
  "resourceOwner": "texas-whiskey"
}

 syncTask.status.failure is triggered on the first failure for that file. When it recovers, data.world will send one syncTask.status.success

Example payload, in JSON:

{
  "datasetid": "sync-demo",
  "event": {
    "type": "FAILURE",
    "task": "state_table.csv",
    "created": "2021-02-04T06:35:17.945Z",
    "error": "HTTP/1.1 404 Not Found"
  },
  "eventType": "syncTask.status.failure",
  "resourceUrl": "https://data.world/texas-whiskey/sync-demo/file/state_table.csv",
  "resourceType": "URL",
  "resourceOwner": "texas-whiskey"
}

Windowed aggregations (beta)

Have you ever need to create a 7-day trailing average in order to smooth out a graph? How about compare a value to the same value in the previous row?

Windowed aggregations are a powerful way to perform complex analysis, such as this, without leaving the comfort of the SQL. These functions operate on a defined group, or "window", of rows to return an aggregated value for each row. Window aggregations are supported for both uploaded tabular data, as well as live tables (where support exists in the connected database).

Note: this feature is currently in Beta. Please contact support with any issues.


New resource types for autosuggest search bar

Find what you are looking for from anywhere in the application.
Look for new resource types added to the autosuggest feature of the main search bar! Queries and metadata resources, such as glossary terms and analysis records, have been added to the top level search bar. You can now search for these resources from any page with the search bar header. This feature previously supported dataset, project, and user search. This feature is live for community and enterprise users.


January 2021 monthly product update

Here is a roundup of some of the key features we released last month, December 2020 -- collection crowdsourced suggestions, new metadata and search API endpoints, and a recap on search improvements.

Coming Soon: Access and ownership moved to Settings

Look for this new feature in upcoming data.world releases!

Here's a sneak peek of what to expect:

We are updating the way we present contributors and visibility for datasets and projects. Currently, we have a top level Access tab that controls individual access, and a Visibility section in the Settings tab that controls whether the dataset/project is open, private, or discoverable to the community. To minimize confusion and complexity, we are combining and redesigning these settings into a single pane of glass to simplify understanding and control of “who can see this resource."

This is how the experience will look from a data consumer's perspective:

This is now the experience will look from an admin's perspective:

Featured Search Results for Data Partners

Relevant data partners now appear at the top of some search results as featured results, making it easier to find high-quality, reliable datasets in search. Featured results are disabled for private installs.

If you’re interested in learning more about how we help teams find the right data vendors for their projects, drop us a line at concierge@data.world.


Show Previous EntriesShow Previous Entries