Administer > Database administration > Data persistence > IR Expert > How IR Expert evaluates documents for relevance

How IR Expert evaluates documents for relevance

IR Expert queries return results based on relevance to the query. To do this, IR Expert looks at each term used in an IR query and gives a ranking to the term, based on how often it appears in the stored documents.

A term found in many documents has a lower rank than a term found in a few documents. For example, if all of the incidents that customers report involve the Windows operating system, then the term “Window” is in almost every document and has a very small ranking.

After IR Expert assigns ranking, it gives each stored document a weight based on how many of the terms used in the query are in the document, and on how often a term is in the document. A document that contains a term twice has a greater weight than a document that contains the term once. Next, it compares the terms in the document with the terms in the query to see if there is a “phrase” match. If so, IR Expert gives that document a higher weight. Finally, IR Expert considers the most recently updated document to be the most relevant.

Related topics

IR Expert
Chinese, Japanese, and Korean language analyzer
Stop words
Creating an IR file

Access IR Expert
Edit queries for Find Solution
Implement IR searches
Load data files with IR Expert keys
Start IR Asynchronous mode

IR Expert file descriptions
Normals dictionary
Stem dictionary
Stemming example
Stop word file
Suffix file