![]() | ![]() | ![]() | Lexical Entry Structure |
FrameNet lexical entries are sets of lexical sub-entries, each of these being a record of what FrameNet has recorded for the lemma in one of its senses. (Since our work has proceeded one frame at a time, rather than one lemma at a time, there are not many instances of multiple sub-entries in the databank. There are, however, cases in which annotators have noted multiple senses by marking them with sentence-initial sense numbers. Unfortunately, no automatic mechanism currently exists allowing such individually marked senses to be associated with their own frames, definitions and valence descriptions.)
An individual lexical entry, then, covers a lemma in a particular part of speech, e.g., as verb or as noun. A lexical sub-entry is intended to represent a single lexical unit, i.e., a lemma in a given part of speech in a single sense.
A lexical sub-entry comprises the following components:
The Frame Element Realization table and the Valence Pattern table are derived automatically from the sentence annotations. Each item in each of these tables is linked to the annotations that exemplify it.
In short, a FrameNet entry provides information, for each sense, about frame membership and the syntactic means by which each Frame Element is realized in the word's surroundings, and summarizes, as Valence Patterns, the full range of combinatorial possibilities as attested in the Corpus.
![]() | ![]() | ![]() | Lexical Entry Structure |