Struct ChatIntegration 

pub struct ChatIntegration { /* private fields */ }

Integration with oxirs-chat for conversational AI

Implementations

impl ChatIntegration

pub fn new(model: Box<dyn EmbeddingModel>) -> Self

Create new chat integration

pub fn with_context_window(self, window_size: usize) -> Self

Configure context window size

pub fn with_similarity_threshold(self, threshold: f32) -> Self

Configure similarity threshold for relevant entities

pub fn extract_relevant_entities(&self, query: &str) -> Result<Vec<String>>

Extract entities from the model’s vocabulary that appear as a substring of query (case-insensitive).

Rather than scanning the whole query once per known entity (O(entities × query length)), this builds a lowercase-name index once and scans the query a single time for bounded-length substring windows against that index (O(entities + query length × longest entity name)).

pub async fn generate_context_embedding( &self, messages: &[String], ) -> Result<Vector>

Generate a context embedding for a conversation by encoding the most recent messages (bounded by context_window) with the held model and averaging the resulting vectors component-wise.

pub async fn generate_personalized_embedding( &mut self, user_id: &str, query: &str, conversation_history: &[String], ) -> Result<Vector>

Generate personalized embeddings for a user

pub fn update_user_profile( &mut self, user_id: &str, query: &str, response_feedback: Option<f32>, interaction_type: InteractionType, ) -> Result<()>

Update user profile based on interaction

pub async fn translate_query( &self, query: &str, source_lang: &str, target_lang: &str, ) -> Result<String>

Translate query to target language

pub async fn detect_language(&self, text: &str) -> Result<LanguageDetection>

Detect language of input text

pub async fn generate_cross_lingual_embedding( &self, text: &str, source_lang: &str, target_lang: &str, ) -> Result<Vector>

Generate cross-lingual embeddings

pub async fn align_entities_across_languages( &self, entity: &str, source_lang: &str, target_langs: &[String], ) -> Result<HashMap<String, String>>

Get multilingual entity alignment