[sword-devel] Programmatic assigning of sword numbers

[Troy A. Griffitts]

I think you should do it!  :)


We actually have a "Guess at Strong's" button on our KJV2003 tagging 
tool.  It did pretty good and gave taggers a good starting place for 
each verse, and DM, it's in JAVA! :)

http://crosswire.org/sword/kjv2003

Actually, the code is pretty trivial.  It just uses known KJV 
translation phrases and frequencies to weight likely candidates in a 
verse.  And like you have mentioned, since we know all the Strong's for 
a single verse, and a single verse only has so many english words, the 
options are narrowed significantly.

So, are you gonna do it? :)

	-Troy.

DM Smith wrote:
> On Sep 23, 2007, at 11:09 PM, jonathon wrote:
> 
>> DM Smith wrote:
>>
>>> If we had a mapping of which words and or phrases in a given  
>>> language mapped to a Strong's number, then one could write a  
>>> program that tagged best matches.
>> That makes it sound far more easier than it is in practice.
>>
>>> Now I've taken enough language classes to know that the effort would
>>> be imprecise. Proofreading would be necessary.
>> I've been looking at English mappings.
>> The same English word can have two or more Strong's numbers.
>> The same Strong number can have half a dozen English words.
> 
> Yes these are some of the problems. Phrase mapping is yet another set  
> of problems.
> 
> But for a given verse, we have a limited number of words for which  
> the Strong's numbers can map. This reduces the ambiguity and in some  
> cases might eliminate it.
> 
> The Strong's Lexicons sometimes list the possible mappings for a  
> particular translation. If we have a translation that is fairly close  
> to that, then we can use that as a  hint of a constrained mapping.
> 
> For example, we have the mapping for the KJV. We have many variants  
> of the KJV at Sword without Strong's numbers. A comparison of the one  
> to the other can yield the differences and applying an approximation  
> mapping algorithm (bitap, for example) can produce good results.
> 
> See a real working example of this here:
> http://neil.fraser.name/software/diff_match_patch/demo_patch.html
> 
> -- DM
> 
> 
> _______________________________________________
> sword-devel mailing list: sword-devel at crosswire.org
> http://www.crosswire.org/mailman/listinfo/sword-devel
> Instructions to unsubscribe/change your settings at above page