Bitcointalk Search Project - Vod's Version

[JollyGood]

It will be a monumental task putting this together therefore you should consider sending PMs to those you would like to see as part of the endeavour otherwise you will not know whether it is feasible to move forward. If there is consensus between the key personnel you would like to participate then it makes life easier when going forward. If there is little (or zero) consensus, I guess that would be the end of the project.

As always, things have to come down to the finances as well. As far as total (estimated) first year costs are concerned, what sort of numbers did you calculate?

I agree with you.   Adding a responsive search to this forum would uncover a lot of past corruption (from the administration and their staff)  so I doubt the whales here will spend a single satoshi making information more available.   As a moderator told me a few years ago about working for Theymos - "Ignorance is bliss".  But since I've started looking into this I've found much better ways than using a GPU for parsing LLMs.  JJ has proven he is talented (and I'm hoping Ninja gets involved as well) so I think they will be able to run a small search on standard shared hosting.  It may take them more resources to train the model, but once trained/indexed it can be run easily on free resources from cloud providers.  These frameworks include a token purchase/use construct, so the developers can make their money back that way for people using the search for non-casual uses.   In other words, this will be a project for the community, and not another tool under admin control.

[Vod]

It will be a monumental task putting this together therefore you should consider sending PMs to those you would like to see as part of the endeavour otherwise you will not know whether it is feasible to move forward. If there is consensus between the key personnel you would like to participate then it makes life easier when going forward. If there is little (or zero) consensus, I guess that would be the end of the project.

As always, things have to come down to the finances as well. As far as total (estimated) first year costs are concerned, what sort of numbers did you calculate?

I'm not really interested in running this project, for reasons I have stated.  I don't want to spend any more of my personal funds improving the forum when the funds meant to do so provide shelter and support to future felons.  The most important part of this project will be the source of the posts.  Will the search engine be based off the historical posts that LoyceV and Ninja have, or will it be the heavily modified/deleted version that Theymos will provide?  Since he is backed by literally millions of dollars with no oversight, it's foolish to design anything against him, right?

The interest is there - all this project really needs is someone to bring together the talent and provide them a place to work.   Also, determine right away if anyone is planning their own advanced search engine (like Ninja/JJ) as it wouldn't be worth it to compete.

I have a friend who is capable of doing this, i can reach out to him for this. If that is fine by you.

Is your friend here on the forum?  If not, why would he be interested in doing this?  This project is not paid so he would need to have an interest in the subject before he spent much time on it.

[Silentcursor]

Is your friend here on the forum?  If not, why would he be interested in doing this?  This project is not paid so he would need to have an interest in the subject before he spent much time on it.

He is not in the forum. I reached out to him after seeing this post. What i can do is to tell him to join the forum then can communicate directly and also join the project since it is an open source. You will get my feedback soon

[u0u]

We already have the full infrastructure ready to go (our own datacenter, GPU nodes, AI models, scalable backend) and top notch developers, We can turn this into a working product with a user-friendly interface hosted by us and maintained if you can find some funding, Or we can offer it on a subscription model if we see enough demand for it...

[Vod]

We already have the full infrastructure ready to go (our own datacenter, GPU nodes, AI models, scalable backend) and top notch developers, We can turn this into a working product with a user-friendly interface hosted by us and maintained if you can find some funding, Or we can offer it on a subscription model if we see enough demand for it...

Go for it!  Sounds like you do not need any funding, and can recoup the investment through API fees.

I'm releasing a project this month that will allow a user to search bitcointalk easily from anywhere - I just need an engine to attach it to. 

[u0u]

For us to move forward meaningfully, we'll need either a full data dump or at least a large enough representative sample to train the model effectively,
I assume some of you already have a local copy or could potentially request access from the forum admins. If that is something anyone here can assist with, it would move things forward meaningfully (since we currently don't have the time to scrape the forum again from scratch).

Also, to clarify the scope: Should we focus just on topic titles and topic content, or include full threads + replies, discussions, and user context? The latter enables a much more powerful, context-aware search experience, but increases training requirements and cost.
Once we've got that clarified, we can prepare a working demo or prototype to showcase what the engine can do.

[NotATether]

For us to move forward meaningfully, we'll need either a full data dump or at least a large enough representative sample to train the model effectively,
I assume some of you already have a local copy or could potentially request access from the forum admins. If that is something anyone here can assist with, it would move things forward meaningfully (since we currently don't have the time to scrape the forum again from scratch).

I'm going to launch an API for retrieving Bitcointalk posts within the next couple weeks, so stay tuned.

[Vod]

For us to move forward meaningfully, we'll need either a full data dump or at least a large enough representative sample to train the model effectively,
I assume some of you already have a local copy or could potentially request access from the forum admins. If that is something anyone here can assist with, it would move things forward meaningfully (since we currently don't have the time to scrape the forum again from scratch).

Also, to clarify the scope: Should we focus just on topic titles and topic content, or include full threads + replies, discussions, and user context? The latter enables a much more powerful, context-aware search experience, but increases training requirements and cost.
Once we've got that clarified, we can prepare a working demo or prototype to showcase what the engine can do.

I've bolded the part that helps explain your disadvantage here.   You are new to the forum and did not read the discussions we've had over the last month.   There are a few forum post sources available (already discussed) and many users have already commented on what they want to see in the results. 

I'm going to launch an API for retrieving Bitcointalk posts within the next couple weeks, so stay tuned.

Cool - if you mean querying posts it's always good to have multiple search options.  I'd like this thread to stay in the spirit of an "AI assisted" search, as there there are already many topics on SQL queries and APIs.   Right now both you and Ninja are working on APIs and may believe the first one to complete will be the winner, which I feel is false.   The engine itself will not be the winner, but the various projects that use their results.  The project I am working on will have the ability for a user to select their engine, soi please guys, take your time and do it right. 

A proper AI search engine might cost close to a thousand dollars to build correctly (or 0.000013BTC out of the 1000BTC donated for projects like this) so there is no reason any user should have to pay.  The largest obstacle will be Theymos anyway - imagine what people would uncover with a Deepseek interface trained on this forum. 

[u0u]

We have decided to take a different approach and have now developed a working AI Search prototype,
i will have it running live as an experimental demo for the next few days. You're welcome to test it out here:
https://my-zone.net/bitcointalk

Please note that you will need to register an account with your email to access the AI Search, this is just a basic measure to prevent abuse.
it's purely conversational and can understand queries about Bitcointalk.

Let me know what you think after trying the demo, all feedback helps.

Some Screenshots:
https://i.ibb.co/sdDtNWdS/n9fAQqc.png
https://i.ibb.co/5gX5Jkt0/image.png

[TryNinja]

We have decided to take a different approach and have now developed a working AI Search prototype,
i will have it running live as an experimental demo for the next few days. You're welcome to test it out here:
https://my-zone.net/bitcointalk

Please note that you will need to register an account with your email to access the AI Search, this is just a basic measure to prevent abuse.
it's purely conversational and can understand queries about Bitcointalk.

Let me know what you think after trying the demo, all feedback helps.

Looks very cool.

I got a 500 Server Error return when asking the AI "who is TryNinja" just now.

edit: Looks like the entire thing is down, can't even refresh the page.

edit 2: Working again, answered me well.

[u0u]

We have decided to take a different approach and have now developed a working AI Search prototype,
i will have it running live as an experimental demo for the next few days. You're welcome to test it out here:
https://my-zone.net/bitcointalk

Please note that you will need to register an account with your email to access the AI Search, this is just a basic measure to prevent abuse.
it's purely conversational and can understand queries about Bitcointalk.

Let me know what you think after trying the demo, all feedback helps.

Looks very cool.

I got a 500 Server Error return when asking the AI "who is TryNinja" just now.

edit: Looks like the entire thing is down, can't even refresh the page.

edit 2: Working again, answered me well.

Thank you for your feedback!
We’re constantly sending updates and improvements, so the error you encountered was likely due to some modifications being deployed at the time. Everything should be stable now, let us know if you run into anything else!

[Vod]

We have decided to take a different approach

Different approach means different thread please.

[SmartGold01]

We have decided to take a different approach

Different approach means different thread please.

Yes you are right and shouldn't be posted here, instead they may have their thread for their approach.

We have decided to take a different approach and have now developed a working AI Search prototype,
i will have it running live as an experimental demo for the next few days. You're welcome to test it out here:
https://my-zone.net/bitcointalk

Please note that you will need to register an account with your email to access the AI Search, this is just a basic measure to prevent abuse.
it's purely conversational and can understand queries about Bitcointalk.

Let me know what you think after trying the demo, all feedback helps.

Looks very cool.

I got a 500 Server Error return when asking the AI "who is TryNinja" just now.

edit: Looks like the entire thing is down, can't even refresh the page.

edit 2: Working again, answered me well.

Yeah.. I also tried to access it and it's working fine. Though also make some random search it gave me something not that serious. From the op response there will be more advancement after more new data has been sourced.