Search before asking
Description
Following the design proposal in #143 and infrastructure introduced in #321, we want to add a Milvus-based vector store implementation to enhance RAG capabilities with Milvus in Flink Agents. This issue focuses on implementing the Java Milvus vector store integration.
The implementation should include:
- Add
MilvusVectorStore under integrations/vector-stores/milvus
- Reuse the existing
BaseVectorStore, Document, VectorStoreQuery, and VectorStoreQueryResult APIs
- Support Milvus connection configuration, such as
uri, token, username, password, and database
- Support collection/search configuration, such as
collection_name, vector_field, text_field, primary_key_field, output_fields, and search params
- Convert Milvus query results into the common
Document type
- Add unit tests and/or integration tests for query argument parsing and result conversion
Are you willing to submit a PR?
Search before asking
Description
Following the design proposal in #143 and infrastructure introduced in #321, we want to add a Milvus-based vector store implementation to enhance RAG capabilities with Milvus in Flink Agents. This issue focuses on implementing the Java Milvus vector store integration.
The implementation should include:
MilvusVectorStoreunderintegrations/vector-stores/milvusBaseVectorStore,Document,VectorStoreQuery, andVectorStoreQueryResultAPIsuri,token,username,password, anddatabasecollection_name,vector_field,text_field,primary_key_field,output_fields, and search paramsDocumenttypeAre you willing to submit a PR?