A large unstructured data is stored through the analysis of the streaming layer of ELA Core-Engine. By real-time streaming processing and recombination of full stream, data redundancy is prevented and integrity is checked. This is a next-generation architecture that makes it easy to handle large storage processing, allowing each node to send and receive the requested data nearly at real-time by automated analysis and automatic storage. It is also linked to learning and inference n the ELA Analysis System and works in conjunction with the blockchain.
Large strage balancing cluster Real time collection Applying distributed queue Structured/ unstructured real time preprocessing Real time analysis and AI real time detection Various distribution storage Predicting performance and fusibility and proactive management Various UI/UX and dashboard
Preprocessing, Model design, Learning Management and distribution of result, Supportig learning pipeline with entire inference Supporting various machine and deep learning algorithm Strong learning process and visualizing result Supporting real time learning preprocessing Supporting selection of dynamic variables from learning data Supporting emsemble
[ ELAMACHAIN AI Big data Curation ]
ELAMACHAIN provides an AI big data curation. Curation services refers to providing ELAMACHAIN’s API to AI-based start-ups and other businesses in need of AI solution. Thus, AI developing companies and AI service companies pay big data curation fee with ELAC. Big data curation service offers more opportunity for ELAMACHAIN project to maintain, repair, supplement, develop and strategically advance AI consistently as part of its operations. ELAMACHAIN’s AI big data curation tool provides basic data analysis such as AI activation, total user number, active user number, and user interaction as well as information for accuracy and target achievement rate of AI comprehending user intent, and user insights that can show user interests. Morover, by aligning and sorting scattered documents and comprehending collected data in depth, such data can be provided to interested parties using the curation tool.
[ AI Activation Level ]
Most fundamental analysis data, AI activation level service cover data on total users, retention rate, etc, which valuable information on how activated AI or the chatbot is.
[ AI performance ]
Though the goal of AI development is to accurately acknowledge user intent and respond to it, there comes many realistic challenge for AI to achieve such perfection. If an AI can numerically assess it degree of failure to meet user’s original intent for every question it can avoid making same mistakes on similar question and environments. AI still can end up with number of unintended result even when the AI engine is configured for over 80% in KPI (Key Performance Indicator), and that is why quantitative analysis on AI performance works as a valuable step to enhance its ability to respond different circumstances correctly based on objective data results.
[ Goal Completion Rate ( GCR ) ]
GCR works as an index of assessing the goal of AI development itself. If a chatbot’s intent activation level comes out low for services with purpose of “recommending cosmetic surgery packages,” it may be about the time to re-constuct the scenario. Likewise, we also need to check the retention rate if its goal is to make users land on a certain page. For a counseling chatbot, in can check the status of goal completion based on indices like increase in counseling cases.