[seqfan] Re: Linked Open Data (LOD) and the OEIS

Marc LeBrun mlb at well.com
Sat Mar 17 22:07:58 CET 2012


>="Neil Sloane" <njasloane at gmail.com>
>> YES. ANOTHER SET OF RELATIONS INVOLVE GENERATING FUNCTIONS,
> SEE THE "transforms" FILE FOR MANY OTHER EXAMPLES.

Yes, there are multiple levels of abstraction that can, and should, be
supported.  We have knowledge about how sequences are related by transforms,
and there's also knowledge about how transforms relate to eachother (eg
transform f is inverse of transform g) and there's even higher-order
knowledge, such as what inverse transforms are in the first place, for
general f and g, and so on.

>> NO, WE DON'T NEED A NEW KEYWORD. WE JUST PUT CONJECTURES
> LIKE THIS IN THE "Formula" SECTION. THERE ARE ALREADY MANY EXAMPLES OF
> THIS.
...
>> WE JUST EDIT THE TEXT. AGAIN THERE ARE MANY EXAMPLES OF THIS IN THE OEIS.

Just to be clear: I am NOT advocating mucking up the existing OEIS!  Every
sequence already has a unique URL.  This enables the "classic" data schema
to be augmented in a decoupled way via semantic web federation.  This option
should be considered as an alternative to hacking the "native" schema for
implementing proposed extensions and new service functionality.

Also, a lot of technology is available these days for processing "free
text".  It was developed for indexing "wild" sources such as tweets, blogs,
news stories, journal articles, commercial web sites, etc.  This will enable
productive automated mining of the Comments, Formulas, References and other
non-numeric parts of the OEIS, leveraging the existing stylistic discipline
and thoughtful editing without requiring extra effort to make it "machine
friendly".





More information about the SeqFan mailing list