A Search Relevancy Tuning Method Using Expert Results Content Evaluation

TL;DR

This paper introduces an online relevancy tuning method that adjusts word weights based on explicit user feedback to improve search quality within a specific data domain.

Contribution

It proposes a novel method of dynamically modifying word weights using user evaluations, enhancing search relevancy in real-time.

Findings

Words weights adjustment improves search relevance

User feedback effectively guides relevancy tuning

Method shows better results in a specific data domain

Abstract

The article presents an online relevancy tuning method using explicit user feedback. The author developed and tested a method of words' weights modification based on search result evaluation by user. User decides whether the result is useful or not after inspecting the full result content. The experiment proved that the constantly accumulated words weights base leads to better search quality in a specified data domain. The author also suggested future improvements of the method.