An Idea I had

[dwpontius]

Hello,

 

I just joined the forums today and am glad to be here.  (Also, I am not sure whether to post this in a more Technical part of the forum or not; the way I see it, the question deals with Source Documents, so it may as well go here.)

 

I have been toying with an idea regarding Biblical Source Material (such as the Hebrew Bible, LXX, Textus Receptus, etc) for a while now.

 

Does anybody know if it were possible to use Accordance to do the following using automation, search commands, or programming:

1. Access the English definition(s) of each word.
2. Store that information for future use.
3. Use the rules of grammar to, with as few changes as possible, edit that information into being readable by an ordinary person, within the context of each sentence.
4. Write the result somewhere, such as in Notes or Papers.
5. Do this process for each Primary Source Document.
6. Compare and contrast the results.

 

I think it would be neat to have a translation that tries to say in relatively plain English what the various Source Documents say in their respective languages.

 

My gut is telling me that with scripting and a few other tools, it may be possible to do this using some of the features that Apple provides its users, but I would like to see if anybody has ideas about how to do this inside of Accordance.  I have tried to do so manually and found that the process to be vulnerable to user error and also take substantial amounts of time.  Please look at the attached screenshot to see what I mean.

[Brian K. Mitchell]

On 5/21/2024 at 7:21 AM, dwpontius said:

I just joined the forums today and am glad to be here.

Welcome to the forums @dwpontius 

 

 

 

 

 

[Dan Dennison]

On 5/20/2024 at 5:21 PM, dwpontius said:

• Use the rules of grammar to, with as few changes as possible, edit that information into being readable by an ordinary person, within the context of each csentence.

 

This is the challenge. "The rules of grammar" are part rules, and part statistical probability. The grammar textbooks call these exceptions. "Readable by an ordinary person" raises that bar considerably, especially if the sentence requires a lot of rearranging from source to target language. This rearranging is called "alignments" in this 1993 paper: https://aclanthology.org/J93-2003.pdf

 

The rules of grammar method in the literature is called RBMT (rules based machine translation), which is ultimately branded rationalism.  A good overview of the two camps of rationalism and empiricism in natural language processing can be found in this paper, also from 1993: https://aclanthology.org/J93-1001.pdf

 

To "edit that information into being readable" resulted in layered approaches. These were ultimately flattened in the modern end-to-end neural models such as https://research.google/blog/a-neural-network-for-machine-translation-at-production-scale/

 

I think to pass the "readable by an ordinary person" test these days requires a large language model. This is indeed a cool feature request.

 

[dwpontius]

Thank you both for the responses!  Sorry for the delay, I thought that there would be an email notification if anybody responded and was waiting for one.

 

It sounds to me like I underestimated the complexity of the problem.  Perhaps there could be a way to just put the English definition of the words into Notes (or somewhere inside Accordance) where the different meanings each word contains are separated by dashes and grouped with semicolons?  Or at least that is how I would do it.

[Dr. Nathan Parker]

2 hours ago, dwpontius said:

Sorry for the delay, I thought that there would be an email notification if anybody responded and was waiting for one.

 

We do have at least an email digest feature that can go out once daily. I can help you switch it on if need be.

[Ken Simpson]

2 hours ago, dwpontius said:

Thank you both for the responses!  Sorry for the delay, I thought that there would be an email notification if anybody responded and was waiting for one.

 

It sounds to me like I underestimated the complexity of the problem.  Perhaps there could be a way to just put the English definition of the words into Notes (or somewhere inside Accordance) where the different meanings each word contains are separated by dashes and grouped with semicolons?  Or at least that is how I would do it.

Do you mean something like this?

 

[dwpontius]

On 5/31/2024 at 7:57 PM, Ken Simpson said:

Do you mean something like this?

 

That Translation column is the sort of thing I have in mind, yes.

 

Based on what Dennison said, it sounds like part of my idea is not feasible (or at least not easy).

 

In case anybody wants to see what I meant with dashes and semicolons, here is a screenshot of my progress so far.

[dwpontius]

On 5/31/2024 at 7:39 PM, Dr. Nathan Parker said:

 

We do have at least an email digest feature that can go out once daily. I can help you switch it on if need be.

Would it be this setting I found after looking around for a little?

[Dr. Nathan Parker]

I think so. I can double-check mine if need be.