Posted 19 years ago
by tobias weltner
I'd like to suggest another thing regarding language definitions.
Currently, tokens can be identified in two different ways: key and id.
It would be great if tokens could be also assigned custom attributes, maybe a limited number of "category1" to category10" or something like this.
This would make it possible to logically group tokens into specific semantic groups that would ease semantic parsing considerably.
Currently, I work around this by using id as bit mask which works but isn't really beautiful because it makes the identification process slow.
It would already be a great help if there was another property just like id that could be used as bit mask for additional grouping info.
Currently, tokens can be identified in two different ways: key and id.
It would be great if tokens could be also assigned custom attributes, maybe a limited number of "category1" to category10" or something like this.
This would make it possible to logically group tokens into specific semantic groups that would ease semantic parsing considerably.
Currently, I work around this by using id as bit mask which works but isn't really beautiful because it makes the identification process slow.
It would already be a great help if there was another property just like id that could be used as bit mask for additional grouping info.