In 2008, Novamente LLC released the source code for its personal "Novamente Cognition Engine (NCE)," which sparked the development of OpenCog. It is an open-source artificial intelligence framework development project. More than 50 companies, including Huawei and Cisco, use OpenCog. 

OpenCog Prime is a robot and virtual embodied cognition architecture that defines a set of interacting components designed to produce human-equivalent artificial general intelligence (AGI) as a system-wide emergent phenomenon. Ben Goertzel designed OpenCog Prime, and the OpenCog framework is a generic framework for broad-based AGI research.

Components of OpenCog

OpenCog consists of:

• AtomSpace is a graph database that stores "atoms," which are terms, atomic formulas, sentences, and relationships, along with their "values" (valuations or interpretations, which can be of as per-atom key-value databases). A value is something like a truth value. Atoms are unique everywhere, and they can't change. Values, on the other hand, are temporary and can change.
• Atomese is a set of pre-defined atoms used to represent generic knowledge, like conceptual graphs and semantic networks, and to represent and store the rules (in the sense of term rewriting) that manipulate such graphs.
• A group of pre-defined atoms represents a type subsystem, including type constructors and function types. These describe the types of variables, terms, and expressions and the structure of generic graphs with variables.
• A set of pre-defined atoms can write both functional and imperative code. One of these is the lambda abstraction, which turns free variables into bound variables and makes a beta reduction.
• A set of pre-defined atoms make up a satisfiability modulo theories solver. This solver is built into a generic graph query engine and is used to match patterns in graphs and hypergraphs (isomorphic subgraph discovery). This process takes the idea of a structured query language (SQL) and applies it to generic graphical queries. It is a form of graph query language that is more general.
• A general rule engine link rules together. It has a forward chainer and a backward chainer. Rules are precisely the graph queries of the graph query subsystem, so the rule engine is similar to a query planner in a vague way. We can implement it for the different inference engines and reasoning systems, like Bayesian inference or fuzzy logic, or practical tasks, like constraint solvers or motion planners.
• ECAN is the name of a subsystem for allocating attention based on economic theory. This subsystem controls numerous search options during inference and chaining.
• A way to use probabilistic logic networks to make a probabilistic reasoning engine (PLN). The current implementation uses the rule engine to link specific rules of logical inference (like modus ponens) and specific mathematical formulas that give each deduction a probability and a confidence level. This subsystem is like a type of proof assistant that uses a modified version of Bayesian inference.
• The program is called MOSES, which stands for Meta-Optimizing Semantic Evolutionary Search. This program is to find groups of short Atomese programs that do things. You can think of this as a type of decision tree learning that leads to a kind of decision forest or a generalization of a decision tree.
• Link Grammar is a natural language input system using Meaning-Text Theory and Dick Hudson's Word Grammar. It stores semantic and syntactic relationships in Atomese.
• It is a system for making natural language.
• OpenPsi is a way to use Psi-Theory to deal with emotional states, drives, and urges.
• It interfaces with Hanson Robotics robots, such as modelling emotions with OpenPsi. This process includes the Loving AI project, which shows how to meditate.

Conclusion

The main goal is to create virtual humans and three-dimensional avatar characters, similar to other cognitive architectures. To mimic actions like emotions, gestures, and learning is the aim. For instance, the researchers only created the software's emotion module because people have emotions. If artificial general intelligence mimics human intelligence, we can make artificial general intelligence.

Furthermore, OpenCog-based research has been published in peer-reviewed journals and presented at conferences and workshops, including the annual Conference on Artificial General Intelligence. The GNU Affero General Public License governs the distribution of OpenCog. Furthermore, the Artificial General Intelligence Research Institute (AGIRI), the Google Summer of Code initiative, Hanson Robotics, SingularityNET, and other organizations support the ongoing development of OpenCog.