More entities, more accuracy, better results
Better Entity Performance
A new entity recognition model
Our new entity model recognises significantly more entities than before, increasing the number to over 5.6 million. The accuracy of our entity recognition has been improved also, with 95% precision for S&P 500 and Russell 2000 companies. And our new knowledge base means that we now use immutable links for entities.
Increased Accuracy
Enhanced search functionality
Create targeted queries to filter out noise and provide more relevant search results. This includes being able to combine a number of entity parameters that wasn't previously possible, such as sentiment, and entity type, e.g. human, company, etc. You can even search by stock tickers or Wikipedia links for added accuracy.
Dependable Sentiment Analysis
Entity-level sentiment analysis
Increased accuracy for sentiment predictions, conducted at an entity level. Each entity is automatically given an average sentiment polarity prediction based on all mentions in an individual article. Predictions are accompanied with a confidence score on how accurate we think that prediction is.