Struct NlpWindow 

#[non_exhaustive]
pub struct NlpWindow {
    pub tokens: Vec<String>,
    pub analysed_words: usize,
    pub sentences: Vec<WindowSentence>,
    pub words: Vec<WordAnalysis>,
    pub ner: Vec<WordSpan>,
    pub srl: Vec<WindowFrame>,
    pub coref: Vec<Vec<(usize, usize)>>,
    pub entities: Vec<NlpEntity>,
    pub facts: Vec<NlpFact>,
}

One analysis window: a run of whole sentences of up to the model’s word budget, analysed in a single pass.

Windows never cross a paragraph or a turn, and a sentence is never split across two windows. Every index in this struct — heads, spans, mention positions — is a word index into tokens, local to the window.

Fields (Non-exhaustive)

Non-exhaustive structs could have additional fields added in future. Therefore, non-exhaustive structs cannot be constructed in external crates using the traditional Struct { .. } syntax; cannot be matched against without a wildcard ..; and struct update syntax will not work.

tokens: Vec<String>

The window’s words, in order.

analysed_words: usize

How many leading words fit in the encoder’s subword budget.

Words from this index on are reported in tokens but carry no analysis. Equal to tokens.len() for a window that fits.

sentences: Vec<WindowSentence>

The window’s sentences.

words: Vec<WordAnalysis>

Per-word analysis, one entry per analysed word (so words.len() == analysed_words).

ner: Vec<WordSpan>

Named entities as (type, start, end) half-open word spans, using the model’s own NER vocabulary (kniv v7: OntoNotes).

srl: Vec<WindowFrame>

Semantic-role frames, one per predicate — every predicate, not only verbs.

coref: Vec<Vec<(usize, usize)>>

Coreference clusters: each cluster is a list of inclusive (start, end) word spans.

entities: Vec<NlpEntity>

Typed entities, grouped across their mentions. In a conversation the speakers come first, in order of first appearance.

facts: Vec<NlpFact>

Facts between entities, scored against the request schema’s relations.

Trait Implementations

impl Clone for NlpWindow

fn clone(&self) -> NlpWindow

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for NlpWindow

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for NlpWindow

fn default() -> NlpWindow

Returns the “default value” for a type.