Machine Learning Exegesis of Ancient Scripture

Post Reply
gracesalmons
New Member
Posts: 1
Joined: Tue Apr 04, 2023 1:28 pm

Machine Learning Exegesis of Ancient Scripture

#1

Post by gracesalmons »

Hi! My fiancé and I created some pretty powerful code. We used the RoBERTa language model and have posted the code on GitHub. (https://github.com/brownthesr/grp_deep_scripture)
With the code, one can:

Find most similar semantic scriptures and conference talks (using dot product cosine similarity scores). This could be used as an alternative search algorithm or to make a newer topical guide.
effectively cluster according to the author and chapter even with nonlinear relationships.
watch the thematic progression of chapters within the same book.
There is implemented a context window (so it won’t take things out of context)
We also fine-tuned the model on modern translations of scripture to lessen the gap between scriptural language and conference talks/common vernacular
We have plans to fine tune model with interpretations of scripture (such as the institute manual) and also to fine tune a generative model on general conference talks and the standard works.

We are looking for ideas on how to better apply this tool for research purposes and also someday to help everyday gospel material users! One idea that we have is to concretely analyze what percent of Christ's sermons were direct quotes of the Old Testament (or beyond that, which books or prophets he quoted most often). Or maybe someday spruce up the gospel library search engine. Does anyone have connections or ideas on who we should reach out to?
Attachments
Scripture_Project (1).pdf
Here is a drafted paper of what we have done.
(526.22 KiB) Downloaded 157 times
Sampfalcon
New Member
Posts: 1
Joined: Sun Jan 28, 2024 3:58 pm

Re: Machine Learning Exegesis of Ancient Scripture

#2

Post by Sampfalcon »

That is so cool. Hope you found the right person to talk to.
Post Reply

Return to “Mobile Applications”