Smoothing document length
Latest Progress
Search our site:
powered by Terrier and

department of computing science at the university of glasgow
university of glasgow

Frequently Asked Questions

What is the term frequency (tf) normalisation?

The tf normalisation models the tf density of the document length. It aims to smooth the effect of document length, which leads the IR system to favour long documents.

What are the free parameters?

Usually, a tf normalisation method is based on the interal of a specific tf density function of document length. Therefore, each normalisation method has a free parameter. Previous research shows that the tuning of the free parameters is crucial for the retrieval performance.

Why Smooth?

Smooth is proposed to supplement Terrier. It stands for the purpose of the tf normalisation methods, which is to smooth the document length effect. A Smooth is also a type of terrier - which is in theme with the Terrier dog family.


Last Update:27 November, 2004