anno/linking/
kb.rs

1//! Knowledge Base abstraction and multi-KB entity linking.
2//!
3//! # Supported Knowledge Bases
4//!
5//! | KB | URI Namespace | Notes |
6//! |----|---------------|-------|
7//! | **Wikidata** | `http://www.wikidata.org/entity/` | Q-items, most comprehensive |
8//! | **YAGO** | `http://yago-knowledge.org/resource/` | Wikipedia + WordNet + GeoNames |
9//! | **DBpedia** | `http://dbpedia.org/resource/` | Wikipedia infobox extraction |
10//! | **Wikipedia** | `https://en.wikipedia.org/wiki/` | Direct article links |
11//! | **Freebase** | `http://rdf.freebase.com/ns/` | Legacy, mapped to Wikidata |
12//! | **UMLS** | `https://uts.nlm.nih.gov/uts/umls/concept/` | Biomedical |
13//! | **GeoNames** | `https://sws.geonames.org/` | Geographic entities |
14//!
15//! # URI/IRI Standards
16//!
17//! Entity linking produces **Linked Data** compatible URIs following:
18//! - W3C RDF standards for entity identification
19//! - HTTP URIs for dereferenceable entities
20//! - owl:sameAs links between KBs
21//!
22//! # Modern Entity Linking Methods
23//!
24//! ## BLINK Architecture (Meta AI, 2020)
25//! ```text
26//! ┌──────────────┐     ┌─────────────┐     ┌───────────────┐
27//! │ Bi-Encoder   │────►│ Dense Index │────►│ Cross-Encoder │
28//! │ (BERT)       │     │ (FAISS)     │     │ (Re-ranker)   │
29//! └──────────────┘     └─────────────┘     └───────────────┘
30//! ```
31//!
32//! ## ReFinED (Amazon, 2022)
33//! - End-to-end mention detection + linking
34//! - Fine-grained entity typing
35//! - Zero-shot linking capability
36//!
37//! ## GENRE (Meta AI, 2021)
38//! - Autoregressive entity retrieval
39//! - Generates entity names directly
40//!
41//! # Example
42//!
43//! ```rust
44//! use anno::linking::kb::{KnowledgeBase, UnifiedLinker, EntityURI};
45//!
46//! // Create unified linker with multiple KBs
47//! let linker = UnifiedLinker::builder()
48//!     .add_kb(KnowledgeBase::Wikidata)
49//!     .add_kb(KnowledgeBase::DBpedia)
50//!     .add_kb(KnowledgeBase::YAGO)
51//!     .build();
52//!
53//! // Link and get URIs for all supported KBs
54//! let uris = linker.link_to_uris("Albert Einstein", None);
55//! for uri in &uris {
56//!     println!("{}: {}", uri.kb, uri.uri);
57//! }
58//! ```
59
60use serde::{Deserialize, Serialize};
61use std::collections::HashMap;
62
63// =============================================================================
64// Knowledge Base Enum
65// =============================================================================
66
67/// Supported knowledge bases.
68#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Serialize, Deserialize)]
69pub enum KnowledgeBase {
70    /// Wikidata (Q-items) - Most comprehensive, actively maintained
71    Wikidata,
72    /// YAGO - Wikipedia + WordNet + GeoNames
73    YAGO,
74    /// DBpedia - Wikipedia infobox extraction
75    DBpedia,
76    /// Wikipedia - Direct article links
77    Wikipedia,
78    /// Freebase - Legacy (deprecated 2016, mapped to Wikidata)
79    Freebase,
80    /// UMLS - Unified Medical Language System
81    UMLS,
82    /// GeoNames - Geographic entities
83    GeoNames,
84    /// Schema.org - Structured data vocabulary
85    SchemaOrg,
86    /// OpenCyc - General knowledge
87    OpenCyc,
88    /// Custom KB
89    Custom,
90}
91
92impl KnowledgeBase {
93    /// Get the base URI namespace for this KB.
94    #[must_use]
95    pub fn base_uri(&self) -> &'static str {
96        match self {
97            Self::Wikidata => "http://www.wikidata.org/entity/",
98            Self::YAGO => "http://yago-knowledge.org/resource/",
99            Self::DBpedia => "http://dbpedia.org/resource/",
100            Self::Wikipedia => "https://en.wikipedia.org/wiki/",
101            Self::Freebase => "http://rdf.freebase.com/ns/",
102            Self::UMLS => "https://uts.nlm.nih.gov/uts/umls/concept/",
103            Self::GeoNames => "https://sws.geonames.org/",
104            Self::SchemaOrg => "https://schema.org/",
105            Self::OpenCyc => "http://sw.opencyc.org/concept/",
106            Self::Custom => "",
107        }
108    }
109
110    /// Get the SPARQL endpoint URL (if available).
111    #[must_use]
112    pub fn sparql_endpoint(&self) -> Option<&'static str> {
113        match self {
114            Self::Wikidata => Some("https://query.wikidata.org/sparql"),
115            Self::DBpedia => Some("https://dbpedia.org/sparql"),
116            Self::YAGO => Some("https://yago-knowledge.org/sparql/query"),
117            _ => None,
118        }
119    }
120
121    /// Get the API search endpoint (if available).
122    #[must_use]
123    pub fn search_api(&self) -> Option<&'static str> {
124        match self {
125            Self::Wikidata => Some("https://www.wikidata.org/w/api.php"),
126            Self::Wikipedia => Some("https://en.wikipedia.org/w/api.php"),
127            Self::GeoNames => Some("http://api.geonames.org/searchJSON"),
128            _ => None,
129        }
130    }
131
132    /// Is this KB still actively maintained?
133    #[must_use]
134    pub fn is_active(&self) -> bool {
135        !matches!(self, Self::Freebase | Self::OpenCyc)
136    }
137
138    /// Get the owl:sameAs predicate for cross-KB linking.
139    #[must_use]
140    pub fn same_as_predicate(&self) -> &'static str {
141        "http://www.w3.org/2002/07/owl#sameAs"
142    }
143}
144
145impl std::fmt::Display for KnowledgeBase {
146    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
147        match self {
148            Self::Wikidata => write!(f, "Wikidata"),
149            Self::YAGO => write!(f, "YAGO"),
150            Self::DBpedia => write!(f, "DBpedia"),
151            Self::Wikipedia => write!(f, "Wikipedia"),
152            Self::Freebase => write!(f, "Freebase"),
153            Self::UMLS => write!(f, "UMLS"),
154            Self::GeoNames => write!(f, "GeoNames"),
155            Self::SchemaOrg => write!(f, "Schema.org"),
156            Self::OpenCyc => write!(f, "OpenCyc"),
157            Self::Custom => write!(f, "Custom"),
158        }
159    }
160}
161
162// =============================================================================
163// Entity URI
164// =============================================================================
165
166/// A fully qualified entity URI with metadata.
167#[derive(Debug, Clone, PartialEq, Eq, Hash, Serialize, Deserialize)]
168pub struct EntityURI {
169    /// The knowledge base
170    pub kb: KnowledgeBase,
171    /// Local identifier within the KB (e.g., "Q937" for Wikidata)
172    pub local_id: String,
173    /// Full URI
174    pub uri: String,
175    /// Optional human-readable label
176    pub label: Option<String>,
177}
178
179impl EntityURI {
180    /// Create a new entity URI.
181    pub fn new(kb: KnowledgeBase, local_id: &str) -> Self {
182        let uri = format!("{}{}", kb.base_uri(), local_id);
183        Self {
184            kb,
185            local_id: local_id.to_string(),
186            uri,
187            label: None,
188        }
189    }
190
191    /// Create with label.
192    pub fn with_label(mut self, label: &str) -> Self {
193        self.label = Some(label.to_string());
194        self
195    }
196
197    /// Parse a URI to extract KB and local ID.
198    pub fn parse(uri: &str) -> Option<Self> {
199        // Try each KB's base URI
200        for kb in &[
201            KnowledgeBase::Wikidata,
202            KnowledgeBase::YAGO,
203            KnowledgeBase::DBpedia,
204            KnowledgeBase::Wikipedia,
205            KnowledgeBase::Freebase,
206            KnowledgeBase::UMLS,
207            KnowledgeBase::GeoNames,
208            KnowledgeBase::SchemaOrg,
209        ] {
210            if uri.starts_with(kb.base_uri()) {
211                let local_id = &uri[kb.base_uri().len()..];
212                return Some(Self {
213                    kb: *kb,
214                    local_id: local_id.to_string(),
215                    uri: uri.to_string(),
216                    label: None,
217                });
218            }
219        }
220        None
221    }
222
223    /// Check if this is a Wikidata Q-item.
224    #[must_use]
225    pub fn is_wikidata(&self) -> bool {
226        self.kb == KnowledgeBase::Wikidata && self.local_id.starts_with('Q')
227    }
228
229    /// Get CURIE (Compact URI) format.
230    #[must_use]
231    pub fn to_curie(&self) -> String {
232        let prefix = match self.kb {
233            KnowledgeBase::Wikidata => "wd",
234            KnowledgeBase::YAGO => "yago",
235            KnowledgeBase::DBpedia => "dbr",
236            KnowledgeBase::Wikipedia => "wp",
237            KnowledgeBase::Freebase => "fb",
238            KnowledgeBase::UMLS => "umls",
239            KnowledgeBase::GeoNames => "gn",
240            KnowledgeBase::SchemaOrg => "schema",
241            KnowledgeBase::OpenCyc => "cyc",
242            KnowledgeBase::Custom => "custom",
243        };
244        format!("{}:{}", prefix, self.local_id)
245    }
246}
247
248// =============================================================================
249// Cross-KB Mappings
250// =============================================================================
251
252/// Cross-KB entity mappings.
253///
254/// Maps entities between different knowledge bases using owl:sameAs relationships.
255#[derive(Debug, Clone, Default)]
256pub struct CrossKBMapper {
257    /// Wikidata to other KBs
258    wikidata_mappings: HashMap<String, Vec<EntityURI>>,
259    /// Other KB to Wikidata (for reverse lookup)
260    reverse_mappings: HashMap<String, String>, // URI -> Wikidata QID
261}
262
263impl CrossKBMapper {
264    /// Create a new mapper.
265    pub fn new() -> Self {
266        Self::default()
267    }
268
269    /// Add a mapping from Wikidata to another KB.
270    pub fn add_mapping(&mut self, wikidata_qid: &str, other_uri: EntityURI) {
271        self.reverse_mappings
272            .insert(other_uri.uri.clone(), wikidata_qid.to_string());
273        self.wikidata_mappings
274            .entry(wikidata_qid.to_string())
275            .or_default()
276            .push(other_uri);
277    }
278
279    /// Get all URIs for a Wikidata entity.
280    pub fn get_uris(&self, wikidata_qid: &str) -> Vec<&EntityURI> {
281        self.wikidata_mappings
282            .get(wikidata_qid)
283            .map(|v| v.iter().collect())
284            .unwrap_or_default()
285    }
286
287    /// Find Wikidata QID from any other KB URI.
288    pub fn to_wikidata(&self, uri: &str) -> Option<&str> {
289        self.reverse_mappings.get(uri).map(|s| s.as_str())
290    }
291
292    /// Create mapper with well-known entity mappings.
293    #[must_use]
294    pub fn with_common_mappings() -> Self {
295        let mut mapper = Self::new();
296
297        // Albert Einstein
298        mapper.add_mapping(
299            "Q937",
300            EntityURI::new(KnowledgeBase::DBpedia, "Albert_Einstein"),
301        );
302        mapper.add_mapping(
303            "Q937",
304            EntityURI::new(KnowledgeBase::YAGO, "Albert_Einstein"),
305        );
306        mapper.add_mapping("Q937", EntityURI::new(KnowledgeBase::Freebase, "m.0jcx"));
307        mapper.add_mapping(
308            "Q937",
309            EntityURI::new(KnowledgeBase::Wikipedia, "Albert_Einstein"),
310        );
311
312        // Apple Inc.
313        mapper.add_mapping("Q312", EntityURI::new(KnowledgeBase::DBpedia, "Apple_Inc."));
314        mapper.add_mapping("Q312", EntityURI::new(KnowledgeBase::YAGO, "Apple_Inc."));
315        mapper.add_mapping("Q312", EntityURI::new(KnowledgeBase::Freebase, "m.0k8z"));
316
317        // New York City
318        mapper.add_mapping(
319            "Q60",
320            EntityURI::new(KnowledgeBase::DBpedia, "New_York_City"),
321        );
322        mapper.add_mapping("Q60", EntityURI::new(KnowledgeBase::YAGO, "New_York_City"));
323        mapper.add_mapping("Q60", EntityURI::new(KnowledgeBase::GeoNames, "5128581"));
324
325        // United States
326        mapper.add_mapping(
327            "Q30",
328            EntityURI::new(KnowledgeBase::DBpedia, "United_States"),
329        );
330        mapper.add_mapping("Q30", EntityURI::new(KnowledgeBase::YAGO, "United_States"));
331        mapper.add_mapping("Q30", EntityURI::new(KnowledgeBase::GeoNames, "6252001"));
332
333        mapper
334    }
335}
336
337// =============================================================================
338// YAGO-Specific Types
339// =============================================================================
340
341/// YAGO entity with taxonomy information.
342#[derive(Debug, Clone, Serialize, Deserialize)]
343pub struct YAGOEntity {
344    /// YAGO identifier
345    pub yago_id: String,
346    /// Primary label
347    pub label: String,
348    /// YAGO types (from YAGO taxonomy)
349    pub types: Vec<String>,
350    /// WordNet synset (if available)
351    pub wordnet_synset: Option<String>,
352    /// GeoNames ID (if geographic)
353    pub geonames_id: Option<String>,
354    /// Wikidata QID (if mapped)
355    pub wikidata_qid: Option<String>,
356    /// Wikipedia article
357    pub wikipedia_article: Option<String>,
358}
359
360impl YAGOEntity {
361    /// Create a new YAGO entity.
362    pub fn new(yago_id: &str, label: &str) -> Self {
363        Self {
364            yago_id: yago_id.to_string(),
365            label: label.to_string(),
366            types: Vec::new(),
367            wordnet_synset: None,
368            geonames_id: None,
369            wikidata_qid: None,
370            wikipedia_article: None,
371        }
372    }
373
374    /// Get the YAGO URI.
375    #[must_use]
376    pub fn uri(&self) -> String {
377        format!("{}{}", KnowledgeBase::YAGO.base_uri(), self.yago_id)
378    }
379
380    /// Check if this is a person.
381    #[must_use]
382    pub fn is_person(&self) -> bool {
383        self.types
384            .iter()
385            .any(|t| t.contains("person") || t.contains("human") || t.contains("wordnet_person"))
386    }
387
388    /// Check if this is a location.
389    #[must_use]
390    pub fn is_location(&self) -> bool {
391        self.types.iter().any(|t| {
392            t.contains("location")
393                || t.contains("place")
394                || t.contains("city")
395                || t.contains("country")
396        }) || self.geonames_id.is_some()
397    }
398
399    /// Check if this is an organization.
400    #[must_use]
401    pub fn is_organization(&self) -> bool {
402        self.types.iter().any(|t| {
403            t.contains("organization") || t.contains("company") || t.contains("institution")
404        })
405    }
406}
407
408// =============================================================================
409// Modern Neural Linking Methods
410// =============================================================================
411
412/// Configuration for neural entity linking.
413#[derive(Debug, Clone)]
414pub struct NeuralLinkingConfig {
415    /// Model architecture
416    pub architecture: NeuralArchitecture,
417    /// Maximum candidates to retrieve
418    pub max_candidates: usize,
419    /// Use cross-encoder re-ranking
420    pub use_reranker: bool,
421    /// Confidence threshold
422    pub confidence_threshold: f64,
423    /// Enable zero-shot linking
424    pub zero_shot: bool,
425}
426
427impl Default for NeuralLinkingConfig {
428    fn default() -> Self {
429        Self {
430            architecture: NeuralArchitecture::BiEncoder,
431            max_candidates: 64,
432            use_reranker: true,
433            confidence_threshold: 0.5,
434            zero_shot: false,
435        }
436    }
437}
438
439/// Neural entity linking architectures.
440#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
441pub enum NeuralArchitecture {
442    /// BLINK-style bi-encoder (Meta AI, 2020)
443    BiEncoder,
444    /// Cross-encoder only (slower but more accurate)
445    CrossEncoder,
446    /// ReFinED end-to-end (Amazon, 2022)
447    ReFinED,
448    /// GENRE autoregressive (Meta AI, 2021)
449    GENRE,
450    /// mGENRE multilingual variant
451    MGENRE,
452    /// EntQA question-answering style
453    EntQA,
454}
455
456impl NeuralArchitecture {
457    /// Get the HuggingFace model ID for this architecture.
458    #[must_use]
459    pub fn model_id(&self) -> &'static str {
460        match self {
461            Self::BiEncoder => "facebook/blink-biencoder",
462            Self::CrossEncoder => "facebook/blink-crossencoder",
463            Self::ReFinED => "amazon/refined-wikipedia",
464            Self::GENRE => "facebook/genre-kilt",
465            Self::MGENRE => "facebook/mgenre-wiki",
466            Self::EntQA => "EntQA/entqa",
467        }
468    }
469
470    /// Does this architecture support zero-shot linking?
471    #[must_use]
472    pub fn supports_zero_shot(&self) -> bool {
473        matches!(self, Self::ReFinED | Self::GENRE | Self::MGENRE)
474    }
475}
476
477// =============================================================================
478// Unified Linker
479// =============================================================================
480
481/// Unified entity linker supporting multiple KBs.
482#[derive(Debug, Clone, Default)]
483pub struct UnifiedLinker {
484    /// Enabled knowledge bases
485    enabled_kbs: Vec<KnowledgeBase>,
486    /// Cross-KB mapper
487    mapper: CrossKBMapper,
488    /// Entity dictionary (for offline linking)
489    dictionary: HashMap<String, Vec<EntityURI>>, // mention -> URIs
490    /// Neural config (if using neural linking)
491    #[allow(dead_code)] // Reserved for neural linking integration
492    neural_config: Option<NeuralLinkingConfig>,
493}
494
495impl UnifiedLinker {
496    /// Create a builder.
497    pub fn builder() -> UnifiedLinkerBuilder {
498        UnifiedLinkerBuilder::default()
499    }
500
501    /// Link a mention and return URIs for all enabled KBs.
502    pub fn link_to_uris(&self, mention: &str, _entity_type: Option<&str>) -> Vec<EntityURI> {
503        let mention_lower = mention.to_lowercase();
504
505        // Check dictionary first
506        if let Some(uris) = self.dictionary.get(&mention_lower) {
507            return uris
508                .iter()
509                .filter(|u| self.enabled_kbs.contains(&u.kb))
510                .cloned()
511                .collect();
512        }
513
514        // If we have a Wikidata match, expand to other KBs via mapper
515        // (In a real implementation, this would call the appropriate APIs)
516        Vec::new()
517    }
518
519    /// Link and return the primary (Wikidata) URI.
520    pub fn link_primary(&self, mention: &str, entity_type: Option<&str>) -> Option<EntityURI> {
521        self.link_to_uris(mention, entity_type)
522            .into_iter()
523            .find(|u| u.kb == KnowledgeBase::Wikidata)
524    }
525
526    /// Get all URIs for a known Wikidata QID.
527    pub fn expand_wikidata(&self, qid: &str) -> Vec<EntityURI> {
528        let mut uris = vec![EntityURI::new(KnowledgeBase::Wikidata, qid)];
529        uris.extend(self.mapper.get_uris(qid).iter().cloned().cloned());
530        uris
531    }
532}
533
534/// Builder for UnifiedLinker.
535#[derive(Debug, Clone, Default)]
536pub struct UnifiedLinkerBuilder {
537    enabled_kbs: Vec<KnowledgeBase>,
538    use_common_mappings: bool,
539    neural_config: Option<NeuralLinkingConfig>,
540}
541
542impl UnifiedLinkerBuilder {
543    /// Add a knowledge base.
544    pub fn add_kb(mut self, kb: KnowledgeBase) -> Self {
545        if !self.enabled_kbs.contains(&kb) {
546            self.enabled_kbs.push(kb);
547        }
548        self
549    }
550
551    /// Use common entity mappings.
552    pub fn with_common_mappings(mut self) -> Self {
553        self.use_common_mappings = true;
554        self
555    }
556
557    /// Set neural linking config.
558    pub fn with_neural_config(mut self, config: NeuralLinkingConfig) -> Self {
559        self.neural_config = Some(config);
560        self
561    }
562
563    /// Build the linker.
564    pub fn build(self) -> UnifiedLinker {
565        let mapper = if self.use_common_mappings {
566            CrossKBMapper::with_common_mappings()
567        } else {
568            CrossKBMapper::new()
569        };
570
571        let enabled_kbs = if self.enabled_kbs.is_empty() {
572            vec![KnowledgeBase::Wikidata] // Default to Wikidata
573        } else {
574            self.enabled_kbs
575        };
576
577        UnifiedLinker {
578            enabled_kbs,
579            mapper,
580            dictionary: HashMap::new(),
581            neural_config: self.neural_config,
582        }
583    }
584}
585
586// =============================================================================
587// NIL Clustering
588// =============================================================================
589
590/// Cluster of NIL (unlinkable) entities.
591///
592/// When mentions can't be linked to any KB, we cluster them
593/// to track potentially new entities.
594#[derive(Debug, Clone, Serialize, Deserialize)]
595pub struct NILCluster {
596    /// Cluster ID
597    pub id: u64,
598    /// Canonical surface form
599    pub canonical: String,
600    /// All surface forms
601    pub surfaces: Vec<String>,
602    /// Inferred entity type
603    pub entity_type: Option<String>,
604    /// Cluster confidence
605    pub confidence: f64,
606    /// Number of mentions
607    pub mention_count: usize,
608}
609
610impl NILCluster {
611    /// Create a new NIL cluster.
612    pub fn new(id: u64, canonical: &str) -> Self {
613        Self {
614            id,
615            canonical: canonical.to_string(),
616            surfaces: vec![canonical.to_string()],
617            entity_type: None,
618            confidence: 1.0,
619            mention_count: 1,
620        }
621    }
622
623    /// Generate a temporary URI for this NIL cluster.
624    #[must_use]
625    pub fn temp_uri(&self) -> String {
626        format!("urn:nil:cluster:{}", self.id)
627    }
628}
629
630// =============================================================================
631// Tests
632// =============================================================================
633
634#[cfg(test)]
635mod tests {
636    use super::*;
637
638    #[test]
639    fn test_kb_uris() {
640        assert_eq!(
641            KnowledgeBase::Wikidata.base_uri(),
642            "http://www.wikidata.org/entity/"
643        );
644        assert_eq!(
645            KnowledgeBase::YAGO.base_uri(),
646            "http://yago-knowledge.org/resource/"
647        );
648        assert_eq!(
649            KnowledgeBase::DBpedia.base_uri(),
650            "http://dbpedia.org/resource/"
651        );
652    }
653
654    #[test]
655    fn test_entity_uri() {
656        let uri = EntityURI::new(KnowledgeBase::Wikidata, "Q937");
657        assert_eq!(uri.uri, "http://www.wikidata.org/entity/Q937");
658        assert_eq!(uri.to_curie(), "wd:Q937");
659        assert!(uri.is_wikidata());
660    }
661
662    #[test]
663    fn test_uri_parsing() {
664        let uri = EntityURI::parse("http://www.wikidata.org/entity/Q937");
665        assert!(uri.is_some());
666        let uri = uri.unwrap();
667        assert_eq!(uri.kb, KnowledgeBase::Wikidata);
668        assert_eq!(uri.local_id, "Q937");
669
670        let dbpedia = EntityURI::parse("http://dbpedia.org/resource/Albert_Einstein");
671        assert!(dbpedia.is_some());
672        assert_eq!(dbpedia.unwrap().kb, KnowledgeBase::DBpedia);
673    }
674
675    #[test]
676    fn test_cross_kb_mapper() {
677        let mapper = CrossKBMapper::with_common_mappings();
678
679        // Should have mappings for Einstein
680        let uris = mapper.get_uris("Q937");
681        assert!(!uris.is_empty());
682
683        // Should include DBpedia
684        assert!(uris.iter().any(|u| u.kb == KnowledgeBase::DBpedia));
685    }
686
687    #[test]
688    fn test_yago_entity() {
689        let mut entity = YAGOEntity::new("Albert_Einstein", "Albert Einstein");
690        entity.types.push("wordnet_person_100007846".to_string());
691        entity.wikidata_qid = Some("Q937".to_string());
692
693        assert!(entity.is_person());
694        assert!(!entity.is_location());
695        assert_eq!(
696            entity.uri(),
697            "http://yago-knowledge.org/resource/Albert_Einstein"
698        );
699    }
700
701    #[test]
702    fn test_unified_linker() {
703        let linker = UnifiedLinker::builder()
704            .add_kb(KnowledgeBase::Wikidata)
705            .add_kb(KnowledgeBase::DBpedia)
706            .with_common_mappings()
707            .build();
708
709        // Expand known QID
710        let uris = linker.expand_wikidata("Q937");
711        assert!(!uris.is_empty());
712        assert!(uris.iter().any(|u| u.kb == KnowledgeBase::Wikidata));
713    }
714
715    #[test]
716    fn test_neural_architecture() {
717        assert!(NeuralArchitecture::GENRE.supports_zero_shot());
718        assert!(!NeuralArchitecture::BiEncoder.supports_zero_shot());
719    }
720
721    #[test]
722    fn test_nil_cluster() {
723        let cluster = NILCluster::new(1, "John Doe");
724        assert!(cluster.temp_uri().starts_with("urn:nil:"));
725    }
726}
anno/linking/kb.rs

anno/linking/
kb.rs
