Home - Gistneuro/intro GitHub Wiki

Fundamental research projects in the field of behavioral economics have shown that decisions made by market participants are often irrational and influenced by psychological, cognitive, emotional and social factors. Several Nobel Prizes in economics were awarded for the research in this field of economics (Daniel Kahneman (2002), Robert Schiller (2013), Richard Thaler (2017)). Excessive or insufficient response to news information is one of the reasons of the investors irrational behavior. This problem is particularly acute in the crypto assets market, as its high volatility is largely caused by an information background determinant.

Development of information technologies and Artificial Intelligence has made it possible to analyze information from various information sources automatically. A number of recent studies have revealed a correlation between dynamics of prices on stock markets and the content of news and social media resources (R. Peterson “Trading on Sentiment: The Power of Minds over Markets”, P. Tetlock “The Role of Media in Finance”, S. Yang and S. Mo “Social Media and News Sentiment Analysis for Advanced Investment Strategies”). GIST platform is a tool for making investment decisions in the world of based on the analysis and evaluation of information flow with the application of Artificial Intelligence. GIST platform employs the latest Big Data and Machine Learning technologies to automate the following tasks: ●collection of information from an unlimited number of sources; ●content and sentiment analysis;
●filtering of information according by its relevance to the crypto assets market;
●clusterization of information by topics and news events; ●ranking of information and sources by their impact on the crypto assets market; ●evaluation of the actual information background for making investment decisions.

A non-public test version has been launched. English is used for search and recognition . GIST platform continuously seaches for all messages related to the concept of crypto assets (also finding the original source of reposted messages). It automatically tracks a variety of public and private sources of information ( media, information portals, forums, chats, channels, Twitter, social networks) in order to identify and process messages related to cryptocurrency, ICO, blockchain, tokens, etc. GIST platform uses state-of-the-art approaches of computer linguistics (syntactical, semantic and morphological text analyses; domain ontology; sentiment analysis; topic modeling) and deep learning methods (recurrent neural networks) for automatic analysis of the collected information. By using graph algorithms, GIST ranks messages and information sources according to their importance and impact on the market, providing an option to specify the topic of interest (for instance, a certain coin or token).