Package org.machanism.machai.bindex.core
package org.machanism.machai.bindex.core
Supplies persistence, semantic search, and recommendation services for Bindex
metadata.
A Bindex describes a library or integration, including its identity,
version, description, dependencies, and Classification metadata. The types in this
package form the core workflow around that description:
BindexRepositorydefines operations for saving Bindex documents, retrieving a document by its identifier, and finding relevant documents from classification filters and an embedding vector.MongoBindexRepositoryprovides the repository implementation using MongoDB. It stores the serialized Bindex together with searchable classification fields and the classification embedding, and also exposes document deletion and registration-identifier lookup.Pickerturns a natural-language request into one or more classifications with a configured GenAI provider, creates an embedding, and delegates semantic matching to aBindexRepository. It also registers Bindex entries and recursively resolves their dependencies.BindexInfois the compact result model used for recommendations; it contains the selected Bindex identifier, version, description, and similarity score.
A typical integration configures a Configurator, constructs a
MongoBindexRepository with it, and supplies that repository to a
Picker. The configurator selects the MongoDB connection and the GenAI
and embedding providers. Callers should close the MongoDB repository when it
is no longer needed so that its client resources are released.
For example:
Configurator config = ...; BindexRepository repository = new MongoBindexRepository(config); Picker picker = new Picker(repository, config); Collection<BindexInfo> matches = picker.pick(request, 20, 0.75, config);
-
ClassDescriptionRepresents metadata information for a Bindex record, including its identifier, version, description, and relevance score.Repository interface for managing
Bindexrecords.MongoDB-backed repository for persisting and retrievingBindexdocuments.Performs Bindex registration, lookup, and semantic retrieval using embeddings and MongoDB.