Text Mining News System - Quantifying Certain Phenomena Effect on the Stock Market Behavior

Stock market prediction is influenced by manyinternal and external factors. One of these factors are the newsarticles and financial reports related to each listed company. This paper describes a system that is able to extract relevantinformation from this type of textual documents, correlate themwith the stock price movement and determine whether ornot a new released news can and in which proportion willinfluence the market behavior. Predefined ontologies are used forclassifying the news articles and automated ontology extractionfor classifying concepts and super - concepts, on an attempt tomake a semantic mining of the text news. The system is basedon a Multi-Agent Architecture that will investigate, extract andcorrelate the textual data message with the price evolution inorder to better determine buy/sell moments, the trend directionand optimize an investment portfolio. In order to validate ourmodel a prototype was developed and applied to the BucharestStock Exchange Market listed companies.