search

arrigonialberto86 / ruby-band

Machine learning and data mining algorithms for JRuby
Other
89 stars 10 forks source link

Naming consistency with respect to the Weka lib #8

Open paulgoetze opened 10 years ago

paulgoetze commented 10 years ago

Are there deeper thoughts behind using singular names for the Weka modules as Weka::Filter, Weka::Clusterer and Weka::Classifier instead of the plural names used in the original weka library Weka::Filters, Weka::Clusterers and Weka::Classifiers? (Whereas they are placed in plural named directories.)

For me, plural forms would make absolutely sense since there are e.g. several classifier modules and classes in the Weka::Classifier module. Consistency might also help in refering to the weka documentation.

Maybe then also the <Part>_utils modules should be nested in a Weka::<Part> module, so their name can be a simple Utils.

What do you think?

arrigonialberto86 commented 10 years ago

I completely agree with these observations, and I welcome your suggestions. I started out GSOC last year by trying to introduce a "new" naming system which could (at least in theory) homogeneously integrate different libraries, so the main point was to create a central 'core' supporting datatypes from different sources. Nevertheless, I found out that users (often coming from Java) prefer to stick to the more conventional Weka naming system :) So far, I have only added the basics of Apache Math, but I am considering further developments. In the meantime, any suggestion or change to the API structure (or DSL) is more than welcome :)