Amir Zeldes
Amir Zeldes
The English GUM model should resolve cases of the type: ``` That Kim came is the problem ``` The csubj dependent should be converted to a valid markable (via depedit?)....
Markables with coordinate mixed entity constituents should have an ambiguous class
It's possible we have uncertain cues about an entity type from morphology or dependencies, but elsewhere in the document we see better evidence. OOV items should be flagged and stochastic...
Some indexical expressions such as 'this year' or 'today' can be coreferent with an actual lexical NP in the text, especially in news texts where the article is dated. A...
When indefinite anaphors are prohibited, this should apply to pronouns as well. In German this applies both to PIS of the kind "einen" (acc. of 'man'), and some other forms...
It should be possible to run xrenner in "NER mode", i.e. to only output entity information without coreference. Possible modes: - only proper name entities - only items in full...
Some collective nouns can take plural agreement via metonymy: the government ... their citizens
Currently `#1:root` only finds nodes that are the root of a graph in all components. Often we want to know what the root of a subgraph is, e.g. just the...
When document names contain periods, the importer assumes that the substring after the last period is an extension, and not part of the document name. In some scenarios, this assumption...
Arguably it should be the prim edge with sec sequence, and the result in 42 should be attached differently.