What method does K2 use to score documents?
K2 uses a method called Term Frequency to score documents. This method attempts to model the probability that a document is about the concept that the query represents. It uses VQL (Verity Query Language), a hierarchical query language with three “levels” of query operators: ACCRUE-class: Notably AND, OR, and ACCRUE. These operators combine the scores of their children according to a formula. PROXIMITY: These operators determine whether individual query terms occur in certain relationships with a document. The actual score of such operators is either closely related to term frequency (notably the SENTENCE and PARAGRAPH operators) or is based on the minimal distance between some set of terms (notably the NEAR and NEAR/N operators). Leaf operators WORD, STEM, WILDCARD, SOUNDEX, TYPO: These operators give a score based on term frequency and, optionally, the length of the document.
