Improved Search
by Jubin JoseSearch accuracy is improved to 80%
- improved from previous 60 %.
Added more content to explore.
Improved open source code, 250 starts so far.
Improved landing page in website.
Getting a constant traffic of < 5 visitors everyday.
20 beta testers added so far.
-
1
Hi @jubinjose, I love the metrics. How do you measure search accuracy?
Wouldn't you be keen to implement a Google like search (for free) and power Aquila Network with that?
-
1
Hi @BediZajkas, thanks for the warm feedback. Right now, search accuracy is measured on SQuAD dataset this way: (number of times correct retrievals fall in top 5 / total search operations)*100.
Yes, I'm implementing a search engine but not a direct Google alternative. Instead, it will be a collection of small search engines (independent & aggregated) maintained by different people.
It's still in its early stage, anyways, why not give it a try.. :)
-
2
Sounds good! "You can't improve what you don't measure" - I'm obsessed with this, so I really like metrics. :))
@jubinjose I'm so curious about these search engines you described! I'm trying to validate our free, self-service product, by looking at pain points, frustrations associated with existing solutions that fellow indie hackers might experience.
Can I drop you a line via DM? I'd be very keen to know more about your search insights! :)
(the product is a fully managed search API, that allows you to power your search engine with AI, to search content/documents by meaning, like Google search - https://relevance.ai/discovery/. )
-
-
