Re: [xml-dev] Have you used AI, machine learning, and/or dataanalytics t

I'd have to say that conversion between formats, especially complicated technical formats is an order of magnitude more difficult than the usual Bayesian type stuff of being able to do automatic classification. In doing automatic classification, even when you have lots and lots of examples in your corpus, you still need to play around with your algorithms to be able to determine the correct precision / recall, and maybe be willing to identify some documents as unclassifiable.

For this reason, with a conversion where you want things to be 100% correct, I do not believe that it would be a good use of time to try to do it.

Cheers,

Bryan Rasmussen

On Sat, Apr 20, 2019 at 5:58 PM Costello, Roger L. <costello@mitre.org> wrote:

Thank you Tim, Rick, Andreas, and Stephen for your excellent replies!

Tim asked about my use case. I have two different XML vocabularies for aeronautical navigation data. I want to convert XML instances of one to equivalent XML instances of the other. Here is a tiny example of the kind of conversion needed for the aeronautical navigation data:

Both XML vocabularies specify data about magnetic variation, such as the magnetic variation at an airport. Here is an XML instance from one vocabulary:

<MAG_VAR>W014404</MAG_VAR>

Here is the equivalent XML instance in the other vocabulary:

<magneticVariation>

<magneticVariationEWT>West</magneticVariationEWT>

<magneticVariationValue>14.7</magneticVariationValue>

</magneticVariation>

If an AI/machine learning/data analytics engine were given lots and lots of examples, would the engine be able to determine the mapping/conversion between the two representations of magnetic variation?

/Roger