[00:16:05] <wm-bb>	 <mahir256> Now that the hackathon is upon us, I'd like to share a possible implementation of renderers for the Abstract Wikipedia that I've been slowly working on.
[00:16:06] <wm-bb>	 <mahir256> Udiron (from the Ladin for "we will hear", an Italian form of udirono "they heard", and the Bengali pronunciation (উদীরণ) of the Sanskrit उदीरण udīraṇa "saying; speaking; communicating") is a prototype natural language generation system using Wikidata lexemes as primitives to construct syntax trees, these trees resembling those in the Universal Dependencies annotation scheme. It forms complex clauses fr
[00:16:07] <wm-bb>	 <mahir256> At the moment, this effort is incredibly preliminary; only the simplest Bengali utterances can be generated, assuming you pull in each lexeme individually—and even among these sorts of utterances there are some inflectional limitations at the moment—but I suspect that as more machinery is built, the top level will become increasingly more abstract, so that an abstract content system, once developed, can properly hook up t
[00:16:09] <wm-bb>	 <mahir256> https://bitbucket.org/mmorshe2/udiron is where you'll find the source; questions, concerns, issues and pull requests are all welcome. This tool requires the tfsl library, which I'm also mirroring at https://bitbucket.org/mmorshe2/tfsl for convenience' sake.
[21:18:16] <wm-bb>	 <mahir256> Here's hoping that the following helps clear up some, if not all, aspects behind the current structure of the code: https://bitbucket.org/mmorshe2/udiron/src/main/details.md (re @mahir256: Now that the hackathon is upon us, I'd like to share a possible implementation of renderers for the Abstract Wikipedia that I've been slowly working on.
[21:18:17] <wm-bb>	 <mahir256> Udiron (from the Ladin for "we will hear", an Italian form of udirono "they heard", and the Bengali pronunciation (উদীরণ) of the Sanskrit उदीरण udīraṇa "saying; speaking; communicating") is a prototype natural language generation system using Wikidata lexemes as primitives to construct syntax trees, these trees resembling those in the Universal Dependencies annotation scheme. It forms complex clauses fr
[21:18:19] <wm-bb>	 <mahir256> At the moment, this effort is incredibly preliminary; only the simplest Bengali utterances can be generated, assuming you pull in each lexeme individually—and even among these sorts of utterances there are some inflectional limitations at the moment—but I suspect that as more machinery is built, the top level will become increasingly more abstract, so that an abstract content system, once developed, can properly hook up t
[21:18:20] <wm-bb>	 <mahir256> https://bitbucket.org/mmorshe2/udiron is where you'll find the source; questions, concerns, issues and pull requests are all welcome. This tool requires the tfsl library, which I'm also mirroring at https://bitbucket.org/mmorshe2/tfsl for convenience' sake.)