Commit Graph

12 Commits

Author SHA1 Message Date
0fbb9238eb Extract more contextual info from the words.
A property dictionary is now only to be considered equal to a word
when it shares at least one group, or neither has groups.
2017-06-04 18:53:17 +02:00
9d49d00688 Work in progress.
* Test
 * More debugging
 * Base concept
2017-05-28 22:39:07 +02:00
aa7bee4c8b Add perform-verb-over-object and (failing) test. 2017-05-24 23:54:58 +02:00
8e304b2a09 Always create the "groups" set for new elements.
This allows a smaller initial knowledge base.
2017-05-24 22:26:30 +02:00
cbeefcf76b Identify group "concepts" on-flight. 2017-05-24 22:05:46 +02:00
2bfe676b2d Integrate the knowledge ASAP.
If we do this before the parsing we can leverage that semantics
in the matching phase.
2017-05-24 22:03:37 +02:00
a99449c04a Add property-has-value example. 2017-05-24 20:42:15 +02:00
d029ecd91d Implication example. 2017-05-24 20:13:42 +02:00
3cfc03373f Use {set} notation for sets. 2017-05-23 23:29:34 +02:00
e6e8146478 Allow learning from unparsed data in tests. 2017-05-23 23:28:01 +02:00
6693b7deb0 Remove need from as_property info.
Probably this can be improved upon if the data is later
analyzed with it's similars.
2017-05-23 23:16:21 +02:00
5297158110 Package as tree_nlu. 2017-05-23 19:04:44 +02:00