Stackoverflow-Analysis icon indicating copy to clipboard operation
Stackoverflow-Analysis copied to clipboard

Feature: Create a Search Engine of Stack OverFlow with GPT embeds

Open Soubeer opened this issue 1 year ago • 7 comments
trafficstars

The problem to a lot of users is they cannot always get to the core solution of their problem by searching randomly on the internet. Getting recommendations to a solution is great and all but it does not lead the user to what he/she actually is looking for. Just like Google works, I can provide a Search Engine for Stack Overflow which can easily relate to the query of the user and provide the most relevant solutions available in it.

Using the year long dataset we can find out the most similar questions including their frequency, time, interaction and helpfulness. Following the trends, we can find out the questions asked and embed them using GPT embeds which includes tokenizing, etc. Converting them to tensors ,we find the similarity ratio and provide the solution to the user.

There is currently no alternative to this that I can think of.

I would like to take this task under GSSOC'24.

What problem is this feature trying to solve?

Users not getting relevant paths or solutions to their questions asked.

How do we know when the feature is complete?

The most relevant solution is available to the user in sequence of importance (if the solution or a similar one exists)

Soubeer avatar May 13 '24 10:05 Soubeer

We both can work on this

SDprogramer avatar May 13 '24 11:05 SDprogramer

Let me know @Soubeer you need to add @SDprogramer ?

sanjay-kv avatar May 14 '24 06:05 sanjay-kv

We both can work on this

@SDprogramer do you mind pinging me on Discord? ID: Incident#9136

Soubeer avatar May 14 '24 12:05 Soubeer

Let me know @Soubeer you need to add @SDprogramer ?

@sanjay-kv yes Sir you can add him and we can do the task together

Soubeer avatar May 15 '24 02:05 Soubeer

This feature aims to solve the problem of users struggling to find relevant solutions on platforms like Stack Overflow. It proposes a search engine that uses a year-long dataset to identify similar questions and provide relevant solutions. The feature is complete when users can easily find the most relevant solution to their query.

shrawani21 avatar May 15 '24 03:05 shrawani21

Let me know @Soubeer you need to add @SDprogramer ?

@sanjay-kv yes Sir you can add him and we can do the task together

Done

sanjay-kv avatar May 15 '24 05:05 sanjay-kv

do we need to add a video of the working mandatorily? the working is present as the output of our task right at the end of the code

Soubeer avatar May 17 '24 16:05 Soubeer