[BUG-UI/UX] Text search + similarity search does not work for Feedback datasets

[frascuchon]

Describe the bug When I apply a text search with the similarity search feature enabled, the text search is not applied to the result list.

To Reproduce Steps to reproduce the behavior:

1. Open the UI and go to an already created dataset with vectors
2. Apply the "Find similar" action
3. Apply a text query in the search bar that should reduce the number of results
4. See no text filter is applied

Screenshots

Environment (please complete the following information):

• OS [e.g. iOS]:
• Browser [e.g. chrome, safari]: Chrome
• Argilla Version [e.g. 1.0.0]: 1.20.0
• ElasticSearch Version [e.g. 7.10.2]: 8.8.2

[dosubot[bot]]

🤖

Hello @frascuchon! 🖐️

I'm Dosu, an AI assistant here to lend a hand while we wait for a human maintainer. I can help troubleshoot bugs, answer queries, and guide you on your journey as a contributor. Remember, no question is too complex or too simple for me! Looking forward to assisting you with your issue.

Thank you for reporting this issue. Based on the information you've provided, it seems like there's a bug in the simultaneous use of text search and similarity search features in Argilla v1.20.0.

The current version of Argilla does support the simultaneous use of text search and similarity search features. This is evident from the following lines in the provided context:

<p>
  Since <code>1.2.0</code> Argilla supports adding vectors to Argilla
  records which can then be used for finding the most similar records to a
  given one.
</p>
<p>
  This feature uses vector or semantic search combined with more traditional
  search (keyword and filter based). Vector search leverages machine
  learning to capture rich semantic features by embedding items (text,
  video, images, etc.) into a vector space, which can be then used to find
  "semantically" similar items.
</p>

These lines indicate that since version 1.2.0, Argilla has supported adding vectors to records for similarity search. This feature uses vector or semantic search in combination with more traditional search methods, such as keyword and filter-based search.

The text search feature in the Argilla codebase is implemented using Vue.js, a popular JavaScript framework for building user interfaces. The search bar is a Vue component defined in the SearchBar.vue file. When the form is submitted, the submit method is called with the current query as an argument. This method blurs the input field (removes focus from it) and emits a submit event with the current query as the payload. This event can be listened to by the parent component to perform the actual search.

However, from your description, it seems like the text filter is not being applied when the "Find similar" action is used. This could be a bug in the implementation of the search feature.

Unfortunately, as a technical support rep, I don't have the ability to create issues or pull requests. I would recommend reaching out to the maintainers of the Argilla repository directly with this information. They should be able to investigate this issue further and provide a fix if necessary.

I hope this information is helpful. If you have any other questions or need further clarification, please don't hesitate to ask.

Sources

• frontend/components/commons/header/help-info/help-info-content/HelpInfoSimilarity.vue
• frontend/components/commons/header/filters/SearchBar.vue

---

This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

If you want to continue the conversation, start your reply with @dosu-bot.

[damianpumar]

Apparently it works, just keep in mind that this feature works only when the use_markdown is set to false.

[damianpumar]

Consuming these endpoints the results are same: (with query text)

(without query text)

http://localhost:3000/api/v1/me/datasets/b2effb08-e0b9-4099-911c-318485db6f86/records/search?include=responses&include=suggestions&offset=0&limit=50&response_status=missing&response_status=draft