Skip to content

Milestones

List view

  • Agents

    No due date
    •2/4 issues closed
    50% complete2 open2 closed

  • Discovery Suggestion Engine (Core DE Functionality)

    Overdue by 1 year(s)
    •
    Due by June 30, 2025
    •0/3 issues closed
    0% complete3 open0 closed

  • User Interaction (UX) and Comparative Analysis

    Overdue by 1 year(s)
    •
    Due by June 30, 2025
    •0/2 issues closed
    0% complete2 open0 closed

  • Knowledge Nexus Wiki Generation

    * Develop scripts to parse the generated Markdown files. * Create a simple static website (e.g., using MkDocs, Jekyll ???) where each core concept/mechanism has a wiki page with hypertext. * Implement hyperlinking: * Use explicitly extracted connections. * Potentially: Add links based on high HDC vector similarity between concepts (cosine similarity/dot product on bipolar vectors). 1. Update Mechanism: Define the process for the monthly refresh (e.g., adding new PDFs, re-processing if templates change). Ben Waldeck iverable: A populated vector database (CNM), corresponding Markdown summaries, and a basic internal navigation Wiki.

    Overdue by 1 year(s)
    •
    Due by June 26, 2025
    0% complete0 open0 closed

  • Nexus Weaver & Conceptual Nexus Model (CNM) - Back-end Build

    * PDF Ingestion & Text Extraction: Read PDFs, extract text using a structured template. * LLM-Powered Extraction: Automate the process of applying the designed LLM templates to each PDF’s text to extract structured information. Handle rate limits, potential errors. * Markdown Generation: Convert LLM output into standardized Markdown files per PDF/section. Store these. * Concept/Mechanism Vectorization using Hyperdimensional Computing Principles (HDC): @Austin Austin Cook * Identify core entities (concepts, mechanisms, methods) from the structured Markdown. * Develop an HDC encoding strategy: * Assign unique high-dimensional base vectors (e.g., random bipolar vectors) to fundamental entities or keywords. * Represent complex concepts/mechanisms by bundling (element-wise sum/majority vote, similar to hdc_train’s prototype generation) the vectors of their components/properties, potentially using permutations for roles/structure (Vector Symbolic Architectures). This captures compositionality. * Store these HDC vectors in the vector database, linked to the source entity and Markdown summary. * Generate standard dense embeddings (e.g., using Sentence Transformers??) for semantic similarity search alongside HDC vectors. * Vector Database Population: Ingest HDC vectors (and optional dense embeddings) along with metadata (entity name, type, source PDF, Markdown summary link) into the chosen vector database.

    Overdue by 1 year(s)
    •
    Due by June 19, 2025
    •0/1 issues closed
    0% complete1 open0 closed

  • Infrastructure Setup

    1. Infrastructure Setup: @Shaggy Shagor Rahman * Fixed storage for PDFs. * Resources for LLM processing and vector database operations. * Merging GitHub repositories. * Continuous Integration / Continuous Development (CI/CD) Pipelines in Github (?) * Output Format: A structured Markdown format incorporating tables, scores (e.g., confidence, relevance), and clear sectioning.

    Overdue by 1 year(s)
    •
    Due by June 26, 2025
    •0/1 issues closed
    0% complete1 open0 closed