How Voyager Calculates Relevancy of Search Results
- Cooper Labenske
When Voyager executes a query against an index, it lists more-relevant items at the top of the index and less-relevant items at the bottom.
Calculating Relevance Scores
Voyager uses the following field-related factors in determining the relative relevance of index items.
tf (Term Frequency)
This is the frequency with which a query term appears in a document. The more times a term appears, the higher the score.idf (Inverse Document Frequency) If a query term has very low frequency across all items in the index, it contributes more to the relevancy score.
coord (Coordination Factor)
Documents that contain more query terms have higher scores.fieldNorm (Field length) Fields with more words in them have lower scores than fields with fewer words.
NOTE: Scores are always normalized so that they fall between 0 and 1.0. That means that scores are a function of a specific query against a specific index, and cannot be compared across different queries or even with the same query using a different index.