SingularityNET (AGIX) Unveils Distributed Atomspace (DAS) for Advanced AI Knowledge Management
SingularityNET (AGIX) has introduced the Distributed Atomspace (DAS), a decentralized knowledge repository aimed at advancing Artificial General Intelligence (AGI) research. This announcement was made during a session celebrating the Alpha release of OpenCog Hyperon.
The Distributed Atomspace extends the Atomspace hypergraph used by OpenCog Hyperon to represent and store knowledge. It acts as a primary knowledge source for AI agents, encapsulating computational results achieved during their execution. DAS is designed to support multiple connections with various AI algorithms, offering a flexible query interface to distributed knowledge bases.
The architecture of DAS consists of key components such as the Traverse Engine, Query Engine, Cache, and AtomDB. These components play crucial roles in the functionality and efficiency of DAS. The Traverse Engine facilitates hypergraph traversal, the Query Engine processes global queries, the Cache layer accelerates queries involving remote DASs, and AtomDB serves as a Data Access Object for storing atoms.
DAS leverages DBMS indexing capabilities and custom indexes, such as the Pattern Inverted Index, to map patterns to occurrences in the knowledge base. The Query Engine’s pattern matching capabilities enable complex queries to be answered, while proper mapping of knowledge bases to Atomspace nodes and links is essential before loading them into DAS.
DAS server deployment follows a Lambda Architecture using OpenFaaS or AWS Lambda, with functions deployed as Docker containers. A significant milestone in DAS development was the integration with MeTTa using Space API, bridging DAS’s knowledge representation capabilities with MeTTa’s dynamic reasoning and pattern-matching strengths.
SingularityNET, founded by Dr. Ben Goertzel, aims to create decentralized, inclusive, and beneficial AGI. The team comprises experts dedicated to various application areas, working towards the development of AGI beyond current limitations in performance and scalability.