This milestone focuses on laying the groundwork for the AI Synthesis Agent. We'll conduct preliminary research into existing knowledge graph tools and techniques for dialogue analysis. The primary goal is to draft a clear plan outlining our proposed approach for extracting and structuring knowledge from conversational data. This involves identifying key technical considerations potential challenges and defining the basic workflow for subsequent development. The plan will also include a preliminary outline of how we intend to integrate with MeTTa and MORK.

Draft Project Plan: A concise document (e.g. 5-10 pages) outlining the high-level technical approach scope and key phases of the project. Initial Research Summary: A brief report summarizing key findings from our preliminary exploration of relevant tools and methods. Preliminary Task List: A basic list of the main development tasks anticipated for Milestone 2.

$10,000 USD

Draft Project Plan is submitted and reviewed, demonstrating a basic understanding of the project's requirements. Initial Research Summary provides sufficient context for the proposed approach. Preliminary Task List is clear enough to initiate the next phase of work.

This milestone involves the initial implementation of the core components required to process raw conversational data and begin populating a basic knowledge graph. We will focus on building the pipeline for ingesting raw transcripts (e.g. text files) and extracting foundational entities (like speaker names and key nouns). The objective is to demonstrate the ability to transform unstructured text into rudimentary graph nodes and edges laying the technical foundation for more complex semantic structuring in future phases.

Working Codebase (Alpha): A functional version-controlled codebase (e.g. on GitHub) that can ingest raw text transcripts. Basic Entity Extractor Module: A Python script or similar module capable of identifying and extracting primary entities from input text. Initial Graph Output: A demonstration of exporting extracted entities and their most basic relationships into a simple graph format (e.g. CSV JSON or a very basic MeTTa-like S-expression) from a small sample of provided data. Setup & Usage Guide: Basic documentation for running the ingestion and extraction process.

$10,000 USD

The provided codebase successfully processes sample text transcripts without errors. The basic entity extractor correctly identifies and extracts at least 70% of pre-defined simple entities (e.g., speaker names) from a test set. The system demonstrably generates a graph output containing the extracted entities and their direct (basic) relationships. The setup guide allows for successful reproduction of the ingestion and output process by a third party.

Building on the core ingestion and basic extraction from Milestone 2 this milestone focuses on refining the knowledge graph and establishing initial integration with the MeTTa language and/or MORK system. We will enhance the extraction capabilities to identify more complex relationships and infer thematic connections from the conversational data. Tools for distilling and detecting inconsistencies within the generated graph will be explored and partially implemented. The primary objective is to demonstrate how the refined knowledge graph can be effectively consumed by or integrated with symbolic AI frameworks like MeTTa/MORK showcasing its utility for basic reasoning tasks as outlined in the RFP.

Enhanced Codebase: The final codebase on GitHub including improved entity and relationship extraction modules and basic graph refinement functionalities (e.g. de-duplication simple contradiction flagging). MeTTa/MORK Compatible Export: A module or script capable of exporting a significant portion of the generated knowledge graph into a format directly consumable by MeTTa expressions or compatible with MORK. Integration Demonstration: A short video or script showcasing the knowledge graph data being imported into MeTTa/MORK and used for a simple query or reasoning task (e.g. multi-hop question answering on specific entities within the graph). Final Report: A summary of the project's achievements performance analysis on a larger dataset (e.g. 20-30 conversations) lessons learned and recommendations for future development. Comprehensive Documentation: Updated developer documentation including APIs usage examples and guidelines for extending the agent.

$10,000 USD

The enhanced agent demonstrably extracts a richer set of entities and relationships from input data compared to Milestone 2, with improved accuracy (e.g., >85% for defined entity types). The system successfully exports generated graph data into a format that can be loaded and processed by MeTTa and/or MORK. The integration demonstration clearly shows the knowledge graph data being utilized within a symbolic AI environment for a basic reasoning query. The final report provides a clear overview of the project's outcomes and future potential. Documentation is thorough enough to enable a third-party developer to understand, use, and extend the agent.