Stock Trend Extraction using Rule-based and Syntactic Feature-based Relationships between Named Entities

Many research topics still debate to predict the trends of a stock in the financial markets. Trend extraction is an important part of the information retrieved from the financial data sources, such as news articles or web pages. For trend extraction on text document, named entities are identified and relations between them are extracted. These trends are extracted from finding relationships between named entities related words for stock data. The relationships of entities in stock news articles have unstructured, time dependency, different word range and length without syntactic structure. Many previous researchers didn’t propose the trend extraction based on named entities and their relationships. This paper proposes rule-based and syntactic feature-based relation extraction method between named entities for the trend extraction. This proposed system extracts trends by finding the relationships between named entities for stock data. The experimental results extract trends from news using relationships within stock related named entities.