Proposal: Reorganize the detailed dataset page with clearer sections

[xuang7]

Hi everyone,

This is a follow-up discussion on the detailed dataset page layout. From the hackathon projects, we saw many creative interface ideas related to workspace design. I think this is a good opportunity for us to discuss broader improvements to the current detailed dataset page.

Currently, many dataset page improvements focus on refining the existing layout, such as improving spacing, adjusting individual components, and so on. While these changes are helpful, we may also want to explore a more organized and modern layout for the detailed dataset page itself.

Right now, dataset information, file management, preview, and settings are mostly presented on the same page. As more dataset-related features are added, the page may become harder to navigate and maintain. A tab-based or section-based layout may help organize the page more clearly while keeping the same level of information available to users.

Proposed direction

We could consider reorganizing the detailed dataset page into clearer top-level sections or tabs, such as:

• Data card: dataset overview, including name, description, owner, tags, and visibility
• Files: file tree and per-file preview
• Metadata: source, schema overview, and statistics
• Settings: dataset management actions and access-related controls

This could make the page easier to navigate, especially as we continue adding more dataset-related features.

The rich file preview from PR #5112 could also be a useful direction to consider for the Files tab.

References:

Current dataset page

Reference: AI-generated sample layout (for discussion only)

Please feel free to share any thoughts, concerns, or suggestions. Thank you.

[chenlica]

The second one looks much better.

[aglinxinyuan]

Please also make sure it's easy to maintain. There were some features that looked fancy but turned out to be really hard to maintain.

[xuang7]

Agreed! Definitely don't want to add too many overly fancy or unnecessary elements. The goal is to keep the design organized and maintainable.

[mengw15]

I agree that the second AI-generated sample looks quite fancy and has some interesting ideas. At the same time, I feel its overall visual style may not fully match Texera’s current UI theme, so we may not want to adopt it directly.

That said, I think it would be a great idea to borrow some of the useful concepts from it and adapt them into our existing design. For example, organizing the information into different tabs could make the page structure clearer, while showing information such as storage used, related workflows that use this dataset can also be interesting.

[xuang7]

Agree!

[carloea2]

Just wondering, should in general apache/texera main focus be to improve UI/UX?

I think there are other issues in the queue, like: "performance", "AI tools", etc.

[xuang7]

I understand that there are other important priorities in the queue, and UI may not be the main focus. At the same time, since datasets/workflows are important components in Texera, I think it would still be valuable to make these interfaces more organized and easier to use.

[chenlica]

@carloea2 Please check my reply to your same question in another post #5130 (reply in thread)

[xuang7]

Thanks for all the inputs!

The AI-generated layout is only meant to serve as a reference, and we do not need to adopt it directly. I agree that we should keep the design consistent with Texera’s current UI theme and make sure it is easy to maintain.

The main goal is to borrow useful ideas, such as tabs or clearer sections, and adapt them into the existing Texera style. Also, please feel free to share any components or information you think would be helpful to include on the dataset page. Thank you!

[chenlica]

Agreed that we want to have new ideas to improve the site, with an implementation that's easy to maintain.

[Xiao-zhen-Liu]

I think the main issue with the current dataset view compared with the proposed AI-generated view is it less about the layout, but more about the fact that it misses some information that might be very useful to the user, e.g., statistics like "Storage Used", "Used in Workflows", the Schema Overview, Used By, Activity, Lineage, etc. Once you add these information, some UI components like tabs become not just "good to have" but necessary.

But each of these new information will take more than just UI changes to maintain. We definitely need a good design on this topic and evaluate how useful these new information will be, instead of blindly adding more stuff.

[xuang7]

Thanks for sharing the concern! I agree that the key point is not just changing the layout, but deciding what information is actually useful and appropriate to show. We should avoid adding too much "fancy" information just to make the page look richer. The design should focus on improving accessibility and usability, and on making the relevant dataset information easier for users to find and understand.

[Nicoleee1108]

Another thing that may also be worth revisiting is the current workflow workspace navigation / menu bar design itself. Right now many controls are packed into a relatively small top toolbar, which can be difficult for first-time users to understand intuitively.

For example, after running a workflow, it is not immediately clear where users should navigate to view results, intermediate outputs, or execution status. A clearer navigation structure, larger visual grouping of actions, or more explicit entry points for execution results could improve discoverability and overall usability for new users.

I think this is somewhat related to the broader dataset page discussion as well, as Texera continues adding more features, improving information hierarchy and navigation may become increasingly important for usability.

[xuang7]

Thanks for bringing this up. I agree that the current workflow toolbar can be a bit confusing, especially for first-time users, since many controls are packed into a small area. Improving this could make the workflow workspace more usable. This sounds like a good topic for a separate discussion/thread if you are interested.

[xuang7]

Thanks everyone for the input on the possible new direction for the dataset page. I am leaning toward separating the current dataset page into different tabs, such as Metadata, File Management, and Settings. This structure would also leave room for adding more features in the future.

I have some initial ideas for the layout of each tab and the information that should be included, while keeping the existing dataset information available. The detailed layout and styling may be adjusted later. As a first step, I plan to refactor the current dataset page into a tab-based structure and start by implementing the Settings tab.

Please feel free to share any comments or suggestions. Thanks!

Initial tab ideas:

Metadata: display dataset metadata information.

File Management: File preview, uploader, version management

Settings: move dataset-level settings into a separate tab.

[chenlica]

I like the idea of using tabs to organize a page. I wonder when a user clicks on a tab, whether we show a separate page or another section within a big page. We can see how other sites design their tabs and do something similar.

[zyratlo]

I am in support of the current design in the screenshots of only changing a section of the page when the user clicks a tab. In my opinion this is the common accepted design, I feel it is too much clicking and visual separation if the user is navigated away to a separate page.

[xuang7]

For the dataset detail page, I think the tabs should work as sections within the same page: the dataset header stays the same, and only the body content changes when switching tabs. It should not be a separate page for each tab.

Many websites use this kind of in-page section/tab structure, sometimes combined with longer scrolling content within each section. This is also closer to what we currently have, but with a clearer information hierarchy.

[chenlica]

Then I also support this "in-page section/tab" idea.

[zyratlo]

I think the initial screenshots look great! There is good distinction between each tab and the contents of each tab are useful and easy for the user to understand.

On the metadata page, I see there are some other information besides metadata, like Views which would be metrics or statistics. If there is enough metrics about the dataset for its own page, we could have a separate tab like Dataset Metrics or Stats. This could include some basic EDA information too if it is useful. If there isn't a lot of information we can keep it combined with the metadata but make it clear that there is other information besides metadata.

[xuang7]

Thanks for the suggestion. I agree that "Metadata" might not be the best name for this tab, since it contains more than just metadata. Maybe "Data Card" would be better.