Salience is a measure of the contextual significance of words and phrases, such as entities, in a document or corpus of documents. Traditional information theory provides a means to measure salience based on absolute or relative term frequency. However, frequency alone is not always a good indicator of an entity’s importance. Rosoka Software uses a proprietary algorithm to determine salience using additional heuristics, such as shared context, that provides a much more accurate measure of the importance of an entity to the document.


A partner of Rosoka Software uses entity salience to prioritize information transfer for a customer with intermittent internet access. Queries for a particular entity will return documents where that entity is more salient first, ensuring that operations have the most critical information available.