:question: Question(s) of the Week: Hypothetically...
# random
l
Question(s) of the Week: Hypothetically speaking, how would your end-users be impacted if the Browse functionality (see attached GIF) was removed? What workflows would be impacted that cannot be supported by our existing Search experience? @big-carpet-38439, @bulky-soccer-26729 and I are excited for your feedback so we can streamline the DataHub user experience! Please leave your comments in the 🧵 and we’ll pick a random DataHub Swag winner next Monday, July 25th blob excited
b
I see browsepaths as a means for users to discover datasets if he/she is just "browsing" without a goal in mind. The search filters works well but the user would need to understand what they mean. Browsepaths are a more simple way to get started. We also use it to "group" datasets in themes that wont work with the current domain model (for instance, "basic training datasets", "for finance", "for HR") but browsepaths allow us to surface the dataset at multiple locations, whereas currently the UI only allows 1 domain per dataset. (I guess it can be replaced with tagging, but discovering existing tags is still an issue) previously we also use it to mimic folder structures for some file type datasets. We still do, but we also use containers. Some users might find the "folder" appearance more intuitive, though. p.s: we created a UI panel for dataset owners to flexibly edit their existing browsepaths, hence we use it as such. I also referred to the long_tailed_companions datasets a lot in the demo site because i needed a "test" dataset to try out some stuff, and it ranked first in the browsepaths😅
g
+1 to another means for discovery. I don't think the search is good enough yet to truly route people to the exact dataset users may be looking for. On the other hand, browse paths is so overwhelming that it's hard to figure out where to even start. One aspect of browse path that I really like is, once I am on an entity page, I am able to click on the subfolder it's located in and see if there's any other tangentially related datasets that might be present there
Another aspect of tables that might be hurting the value of browse paths is that most datasets are <namespace>.<table_name>. For internal use cases, we have <namespace>.<namespace>.....<namespace>.<table_name>:<version>, with which the browse paths has been really useful
b
it is unfortunate that it only shows the breadcrumbs of the 1st browsepath, though.
m
Huge impact, this is a great discovery tool to see the total number of assets and where, easily able to filter down and see.
b
We use the DataHub’s backend, but not the frontend. We have our own frontend and initially we try to have same experience as DataHub. We got some feedback about this navigation (users complaining it requires too many clicks) and removed it. We have now just the search result. To compensate, we are working in extending the recommendation section in the home page to have some other key grouping.
l
Hey folks, this is some seriously helpful feedback - THANK YOU!! It has really informed our thinking of how we can make Browse really sing