Struct KGDataset 

pub struct KGDataset {
    pub triples: Vec<(usize, usize, usize)>,
    pub n_entities: usize,
    pub n_relations: usize,
    pub entity_labels: Vec<String>,
    pub relation_labels: Vec<String>,
}

A collection of (head, relation, tail) triples for a knowledge graph.

Entities and relations are referenced by their integer indices. Use KGDataset::from_str_triples to build a dataset from string labels automatically.

Fields

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

All triples as (head_idx, rel_idx, tail_idx)

n_entities: usize

Number of distinct entities

n_relations: usize

Number of distinct relation types

entity_labels: Vec<String>

Entity index → string label

relation_labels: Vec<String>

Relation index → string label

Implementations

impl KGDataset

pub fn new( triples: Vec<(usize, usize, usize)>, n_entities: usize, n_relations: usize, ) -> Result<Self>

Build a KGDataset from raw integer triples.

Arguments

• triples – Vec of (head, relation, tail) triples.
• n_entities – Total number of entities (must exceed all indices).
• n_relations – Total number of relation types.

pub fn from_str_triples(triples: &[(&str, &str, &str)]) -> Self

Build a dataset from string-labeled triples.

Entity and relation vocabularies are inferred from the data in the order they are first encountered.

pub fn len(&self) -> usize

Return the number of triples.

pub fn is_empty(&self) -> bool

Return true if the dataset contains no triples.

pub fn corrupt_triple( &self, triple: (usize, usize, usize), ) -> (usize, usize, usize)

Randomly corrupt one triple by replacing either the head or the tail with a uniformly sampled entity (excluding the original).

Trait Implementations

impl Clone for KGDataset

fn clone(&self) -> KGDataset

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for KGDataset

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

Formats the value using the given formatter.