SNBA & UPI
Semantic Network Based Architecture & Unified Platform for Innovations
The creation of this architecture is based on more than 15 years of experience accumulated from the development of the system DOCMAN. It was developed from USW Ltd and provides processing of unstructured data with a graph-based organization. In the version DOCMAN©2.0 of the system, the models of the data were hardcoded. In the next version- DOCMAN©2.5 was attempted an expansion of the basehard coded set of models with new ones, created in the environment of the system. The results weren’t satisfying. The analysis showed that the mixed usage of hardcoded models and freely created models of data in the same environment leads to not only non-optimal constructs but also does not allow achieving homogeneous environment for machine presentation of semantics.
That’s how was formulated the task for creating a homogeneous environment in which by equal means were maintained models of data and their instances i.e., hardcoded models are missing. his, on the other hand, evolved into the definition of the task for creating a platform that provides the developmental infrastructure for creating information systems with an explicit presentation of semantics pledged in the concepts, data, and processes maintained by such systems. The means for presenting semantics are described in Semantics representation regarding establishing and maintaining a Semantic Interoperability in the e-Governance’s environment.
The creation of such a defined platform called Unified Platform for Innovations (UPI) took 7-8 years. It went through the natural stages of testing with practical development of UPI-based systems, analysis of results, improvement, and testing over again when creating more complicated and thematically different systems. At the moment, UPI offers functionally completed environment for creating information systems with Semantic Network Based Architecture (SNBA), which offers innovative solutions to two main problems:
“ Maintenance of Knowledge graph as means of presentation of semantics not only in its general meaning but also as concepts, data, and processes, forming the technological functionality of the UPI-environment
“ Introduction and maintenance of internal system Semantic interoperability
The elaboration of UPI-based, SNBA- systems is done with high productivity of the developmental process and sharply declining the possible mistakes. That’s how is achieved a natural reduction of the distribution of their influence on the whole software realization since it is maximumly decentralized.
On the other hand, the UPI-based systems could be defined as „semantically oriented/centralized“, which gives the opportunity to implement elements of autonomous behavior in them, performing processing of SNBA-presented semantics. Thus arises the problem of accumulation of semantics in a domain specific information system, which is essentially a learning process, as a result of which such a system has learned the field of its application. This result is actually creating new areas of knowledge in the Knowledge graph by creating new nodes and arcs in it.
In the terms of SNBA this means creating new objects in the UPI-based system. The semantics of these objects corresponds the semantics pledged in words or expressions of the natural language. This connection between the semantics of the natural language and UPI-maintained semantics in the SNBA-systems created a new innovative direction in the works of INATO Ltd. The basis of the work in this direction is the creation of the so-called Root Language- RL. The presentation of semantics with RL-means is highly ineffective, with which it is getting closer to the proto-languages in the academic environments. But along with this, the RL-representation has two important advantages.
“ There is a possible automatic translation from a controlled version of a natural language in RL and vice versa
“ There is a possible interpretation in the UPI-environment of the semantics, pledged in RL-descriptions, with which is formed the corresponding UPI-presentation of this semantics which for us is a way of realizing a manageable and controllable process of Machine Learning- ML.
So far, the basic rules of RL-grammar have been developed. For the interpretation of RL-presented semantics is developed a mechanism borrowed by the scheme for interpretation of the language FORTH. The UPI-resources for parallel conducting of ML are developed, through dialogical RL-communication with teachers, presented in the UPI-environment as users of the UPI-system.
In parallel with the ML-process is provided also internal processing of the introduced in the UPI-environment semantics. As an element of that processing is developed an algorithm for identification of generic concepts and automatic creation of their models through which the processing of semantics is evolving to a whole new level of abstraction. In process of development is the creation of models for decision making as an automatic semantics-based composition of processes for their presentation, as well as the opposite task- decomposition of decisions.
The results of the works in the RL-heading are the basis for creating systems with a higher level of autonomous behavior. At this stage, there are preparatory activities being performed for the development of such Smart home-systems for loT and RPA-systems and others
While creating such systems there will be used accomplishments of other innovative developments- for example, modules using a simulation of neural networks, modules doing complicated mathematical processes, and others.