Struct QueryAnalysis 

pub struct QueryAnalysis {
    pub focal_entities: Vec<FocalEntity>,
    pub discriminative_modifiers: Vec<Modifier>,
    pub relational_context: Vec<Relation>,
    pub compound_nouns: Vec<String>,
    pub original_query: String,
    pub has_negation: bool,
    pub intent: QueryIntent,
}

Complete linguistic analysis of a query

Fields

focal_entities: Vec<FocalEntity>

Focal entities (nouns) - primary search targets

discriminative_modifiers: Vec<Modifier>

Discriminative modifiers (adjectives) - quality refiners

relational_context: Vec<Relation>

Relational context (verbs) - graph traversal guides

compound_nouns: Vec<String>

Compound nouns detected (e.g., “machine learning”, “neural network”)

original_query: String

Original query text (retained for logging/debugging)

has_negation: bool

True if query contains negation

intent: QueryIntent

Detected query intent for retrieval strategy selection (SHO-D6)

Implementations

impl QueryAnalysis

pub fn total_weight(&self) -> f32

Calculate weighted importance of this query (for ranking)

pub fn all_stems(&self) -> HashSet<String>

Get all stems for efficient matching

pub fn positive_entity_stems(&self) -> Vec<&str>

Get non-negated entity stems (for positive matching)

pub fn negated_entity_stems(&self) -> Vec<&str>

Get negated entity stems (for exclusion)

pub fn to_ic_weights(&self) -> HashMap<String, f32>

Convert analysis to IC weights HashMap for BM25 term boosting

Returns a mapping of lowercase terms to their IC weights:

• Nouns (focal entities): IC_NOUN = 1.5
• Adjectives (modifiers): IC_ADJECTIVE = 0.9
• Verbs (relations): IC_VERB = 0.7

Based on Lioma & Ounis (2006) - nouns carry more information content.

pub fn to_ic_weights_with_yake(&self, use_yake: bool) -> HashMap<String, f32>

Convert analysis to IC weights with optional YAKE boosting

When use_yake=true, extracts discriminative keywords using YAKE algorithm and boosts their weights. This is critical for multi-hop queries where discriminative terms like “sunrise” must outweigh common terms like “Melanie”.

pub fn keyword_discriminativeness(&self) -> (f32, Vec<String>)

Get maximum YAKE keyword discriminativeness score for dynamic weight adjustment

Returns (max_importance, discriminative_keywords) where:

• max_importance: 0.0-1.0, higher means more discriminative keywords found
• discriminative_keywords: keywords with importance > 0.5 (for logging)

Use this to dynamically adjust BM25/vector weights in hybrid search:

• High discriminativeness (>0.6) → boost BM25 weight (keyword matching critical)
• Low discriminativeness (<0.3) → trust vector more (semantic similarity better)

pub fn to_phrase_boosts(&self) -> Vec<(String, f32)>

Get phrase boosts for BM25 exact phrase matching

Returns compound nouns and adjacent noun pairs as phrases with boost weights. Phrase matching significantly improves retrieval for multi-word concepts like “support group”, “machine learning”, “LGBTQ community”.

Trait Implementations

impl Clone for QueryAnalysis

fn clone(&self) -> QueryAnalysis

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for QueryAnalysis

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

Formats the value using the given formatter.