ARCHI – Artificial Cognitive Hybrid Intellect
Our main product is a new generation of AI platform named ARCHI. It is an open Semantic platform designed specifically for processing of digital knowledge, automated decision making and decision support applications.
Main features of ARCHI are:
Openness: no special requirements for the format of underlying data and information
Knowledge mining tools for finding logical patterns in data to help analysts and experts to get insights
High-load: the core of the ARCHI was initially developed for high-load telecom applications targeting tens of thousands of transactions per second, and horizontal scalability for fast capacity growth
Machine Learning and support of Big Data
ARCHI is based on Semantic Technology, Artificial Intelligence, big data, and Machine Learning.
Semantic technology is a new way to build a bridge between human understanding and how things are represented in a computer. In Semantic technology the traditional coding is replaced by definition of data by other data/processes/rules and their interconnection based on fundamental mathematical logic.
Every data is defined through multiple other data with various interconnections. This makes them instantly interconnected, reusable and with universal meaning. Structured data is called information. The knowledge of how to use/process/decode this information (i.e. meaning or "semantics") is represented in d0sl — "Delta-Zero Semantic Language".
Then knowledge is assembled in concepts, datasets, processes and modules specific for each company and industry, and is meaningfully ("semantically") defined in DSL — Domain Specific Language.
Definitions in DSL allow to integrate all IT systems, software and databases of the company, as well as the information from open sources/ web, to represent the knowledge of the company and employees in a form which can be processed by the IT systems and optimize their utilization.
The Artificial Intelligence component of ARCHI is a Limited Memory Cognitive model based on Machine learning and Reinforced Deep learning.
Artificial Intelligence uses Machine Learning (ML) to take semantically defined data and quickly turn it into actionable information.
Reinforced Deep Learning (RPL), as a subset of ML, creates new algorithms based on given feedback from performed actions. It allows to learn from experience and to ensure the ongoing improvement of ARCHI efficiency.
Artificial Intelligence component of ARCHI understands a set of programs and algorithms elaborated by our developers, and the Administration Interface.
AI component of ARCHI includes 3 layers:
IT systems which customer already has / uses (e.g. in-house databases, ERP and CRM systems, etc).
IT systems are integrated through special drivers that we call subject domain models.
ML modules for analysis, classification and structuring of the data. ML layer is built on the industry standard Apache SPARK, Anaconda, and Hadoop for big data processing, with special proprietary components designed specifically for knowledge mining and natural language understanding.
Upper level business logic represented in a form of DSL definitions and logical rules (i.e. semantic models). DLS is built on d0sl which processes semantic models.
Administration Interface of ARCHI consists in following components:
Semantic Toolkit for business analysts allowing to digitize company knowledge in the form of DSL definitions and logical rules (semantic models), test those models on real data and information, and roll back those changes if they are incorrect.
Version Control and Workflow Management module, allowing to manage and control the Life Cycle of the digital knowledge (semantic models): creation, testing, pre-prod execution, deployment into a production.
ML administration and control tools to control and optimize utilization of the hardware, deployment of new ML components, managing big data.
Statistics and Administration Dashboard allowing to control knowledge execution discrepancies and errors, and analyze explanations
Our package includes the training of your IT staff and ongoing support to simplify the modification and the automation of any new process.