{
"version": "1.0",
"last_updated": "2026-01-31",
"description": "OpenCog use-case to repo mapping based on Linas's vision and active development",
"use_cases": {
"llm_enhancement": {
"description": "Enhance LLMs with long-term memory, crisp reasoning, and symbolic recall",
"repos": ["learn", "sensory", "atomspace", "evidence"],
"status": "active_research",
"linas_quote": "The proximal motivation is to endow Claude with offline memory and reasoning skills that operate outside the LLM itself"
},
"language_learning": {
"description": "Unsupervised grammar induction from text corpora",
"repos": ["learn", "link-grammar", "lg-atomese", "generate"],
"status": "mature",
"approach": "Frequentist counting → vector representations comparable to Word2Vec but as symbolic hypergraphs"
},
"evidence_gathering": {
"description": "Hierarchical similarity assessment and story construction from fragmentary data",
"repos": ["evidence", "atomspace", "sensory"],
"status": "early_research",
"examples": ["genealogy", "file integrity", "scientific investigation", "journalism"]
},
"sensorimotor_agents": {
"description": "Autonomous agents with perception, action, and decision-making",
"repos": ["sensory", "rocca", "learn"],
"status": "active_development",
"key_concepts": ["OpenLink", "WriteLink", "Link Grammar connectors for I/O description"]
},
"knowledge_graphs": {
"description": "Store and query complex relational knowledge",
"repos": ["atomspace", "atomspace-rocks", "atomspace-cog", "cogserver"],
"status": "production_ready",
"key_features": ["pattern matching", "graph rewriting", "TruthValues", "Values"]
},
"nlp_pipeline": {
"description": "Parse natural language to semantic representations",
"repos": ["link-grammar", "relex", "lg-atomese", "learn"],
"status": "mature",
"languages": ["English", "Russian", "Thai", "Arabic", "Persian"]
},
"robotics": {
"description": "Robot control with cognitive architecture",
"repos": ["ros-behavior-scripting", "ghost_bridge", "rocca", "spacetime"],
"status": "experimental",
"integrations": ["ROS", "Hanson Robotics", "OpenAI Gym"]
},
"bioinformatics": {
"description": "Genomic and proteomic reasoning",
"repos": ["agi-bio"],
"status": "maintained",
"applications": ["gene expression analysis", "protein interaction networks"]
},
"program_synthesis": {
"description": "Evolutionary search for programs that solve problems",
"repos": ["moses", "asmoses", "miner"],
"status": "legacy_maintained",
"approach": "Meta-Optimizing Semantic Evolutionary Search"
},
"probabilistic_reasoning": {
"description": "Uncertain inference with probabilistic logic",
"repos": ["pln", "ure", "attention"],
"status": "legacy",
"note": "Superseded by trueagi-io repos for Hyperon"
},
"stream_processing": {
"description": "Data flow processing in Atomese",
"repos": ["learn", "sensory", "atomspace"],
"status": "experimental",
"key_concepts": ["FilterLink", "RuleLink", "graph-as-program"]
},
"distributed_storage": {
"description": "Persist and share AtomSpaces across machines",
"repos": ["atomspace-storage", "atomspace-rocks", "atomspace-cog", "atomspace-dht", "atomspace-ipfs"],
"status": "active_development"
}
},
"patterns": {
"hypergraph_pattern_matching": {
"description": "Find subgraphs matching a template with variables",
"atoms": ["MeetLink", "GetLink", "QueryLink", "BindLink", "DualLink"],
"repos": ["atomspace"]
},
"graph_rewriting": {
"description": "Transform graphs based on pattern-action rules",
"atoms": ["BindLink", "RuleLink", "RewriteLink"],
"repos": ["atomspace", "ure"]
},
"connector_assembly": {
"description": "Link Grammar-style assembly of components via typed connectors",
"key_insight": "Same formalism describes: language syntax, circuit wiring, robot I/O, software APIs",
"repos": ["link-grammar", "sensory", "generate"]
},
"sheaves_and_linkages": {
"description": "Weave together jigsaw-like structural pieces",
"theory": "Mathematical sheaves on graphs",
"repos": ["atomspace", "evidence", "learn"]
},
"vector_symbolic_hybrid": {
"description": "Attach vector embeddings to symbolic graph structures",
"key_insight": "High-dimensional vectors on hypergraph nodes/links",
"atoms": ["FloatValue", "Values"],
"repos": ["atomspace", "learn"]
}
},
"technologies": {
"atomese": {
"description": "Graph representation + query language + executable programs",
"analogy": "SQL-as-a-graph, regex-as-a-graph, AST database",
"bindings": ["C++", "Scheme (Guile)", "Python", "Haskell"]
},
"link_grammar": {
"description": "Typed connectors that must link to form valid structures",
"applications": ["NLP parsing", "circuit design", "API description", "robot I/O"],
"key_insight": "Same math describes syntax, wiring, and interfaces"
},
"truth_values": {
"description": "Probabilistic confidence on atoms",
"types": ["SimpleTruthValue", "CountTruthValue", "IndefiniteTruthValue"],
"use": "Uncertain knowledge, Bayesian reasoning"
},
"values": {
"description": "Mutable key-value data attached to any atom",
"types": ["FloatValue", "StringValue", "LinkValue", "FormulaStream"],
"use": "Vectors, embeddings, dynamic data"
}
},
"priority_order": [
"learn",
"sensory",
"atomspace-storage",
"atomspace",
"evidence",
"link-grammar",
"generate",
"guile-dbi",
"cogutil",
"cogserver"
]
}