metamodelAndRAG.py

Table of Contents

1. Introduction
2. "import" statements
3. Functions
4. Checklist of features across Streamlit and LocalWorkflow

1. Introduction

The functionality for metamodel tweaking using RAG has been isolated in this python file for the purpose of creating further modifications when required to this part of the process. This function is present in both Streamlit and LocalWorkflow versions.

2. "import" statements

[To be completed]

3. Functions and classes

class CustomRetriever(BaseRetriever):

• a self-querying CustomRetriver that is utilised to extract data from a list of metamodel definitions and example RDF files

getRAG_metamodels(openai_key, text_value, metamodel_number):

• extracts the metamodel from the CustomRetriever Object
• makes a call to model.py's RAGPrompt function that handles the model request functionality
• Currently, it is very resource consuming on the HuggingFacePipeline in the LocalWorkflow, hence it is preferred not to use this
• WIP: In future, it is intended to implement one of the Vectorstore retrievers available within the LangChain framework, which might require OpenAI key

metamodel_operations(openai_key, text_value, i, metamodels_to_process, llm_selected="OpenAI"):

• creates the interface between the combiendProcess and metamodelAndRAG files
• stores the output from metamodel alignment in the working directory

4. Checklist of features across formats

(TBD = To Be Developed)

Feature	Streamlit	LocalWorkflow	LegalDocML	LegalRuleML
Self-Querying RAG	Yes	Yes	No	Yes
HuggingFaceOption	No	Difficult (resource-intensive)	No	Yes
VectorStore RAG	TBD	TBD	TBD	TBD