🔷 [Epic] Component Search with Enhanced Filters (AI Tags)

[TiffanyGYJ]

What ?

• High Level Goal: The component classification with AI generated Tags is aim for improving the usability of the component search experience including on the search bar and near.org/component page to improve component discoverability to enable developers to more easily find components they can potentially reuse.

• UX Designs:

  (Link to the actual designs or the task(s) to request design resources) Two areas:

  1. AI tags as filters from component search experience
  2. AI tags serve as part of onboarding on near.org/components to trigger further exploration of component in early stage of their onboarding journey

---

Why ?

• Background: From the developer user research we have conducted, developers mainly are used to and prefer two main ways to discover components to use --- from categorized gallery/library and from search. We have the search functionality in place with improved details page to support the second method. However, there are many cases where developers are not sure how to search or the search is limited due to the metadata available, meaning many components are missing descriptions or tags. As a result, a categorized component gallery is called for. However, with over 14k components and majority lack metadata, this is a challenging issue to solve. The team decided to test with AI models to help with this classification.

• Move the Needle: Key success metrics:

  1. Average CTR of components from component search
  2. Usage of these Tags as filters (in percentage of total user using component search)
  3. Number of recurring usage of these new AI tag filters
  4. Average CTR of components on near.org

---

How ?

• Completion Requirements -> the Definition of Done:

  (Describe in the form of User Stories, the complete set of statements that must be true in order for this body of work to be considered complete.)

  1. As a developer, when I search for infinite scroll, I will be able to get a list of component that matches this search term, and it also contains components that tagged with this by AI instead of already in metadata so that the search is more relevant and helpful
  2. As a developer, when I land on near.org/components page, I will be able to see a list of categories (tags generated by AI with marks) to help me filter over 15k components so that I can get a sense of what is available and how might i use this resource
  3. As a developer, I can report when I see my component is tagged by AI in an inaccurate way and provide what I think it should be by adding in more tags which will override the AI tags.

• Suggested Release Milestones

  1. POC phase focusing on Tags available as Filters - requirement 1
  2. near.org/components onboarding experience - Requirement 2
  3. Make amendable - requirement 3

• Go To Market Needs

  1. We want to make sure the new functionality is discoverable, as a result, we will need to make a it visible either through a banner on near.org or somewhere on homepage so that we make sure developers are aware
  2. A blog post will be made for this feature when we reach milestone 3

### MVP
- [ ] https://github.com/near/data-platform/issues
- [ ] https://github.com/near/data-analytics/issues

[TiffanyGYJ]

@jinouyang Let's chat about the metrics evaluation on the success for this. Would love to see what would be a good benchmark to evaluate if this feature/experiment is successful

[github-actions[bot]]

This issue has been automatically closed because it has been inactive for more than 180 days. Please reopen and prioritize this for development if it is essential.