An experiment to test the utility of repeating phrases for information retrieval systems
Kevin Cox · Journal of Information Science · 1994
This paper describes a method of evaluating the utility of repeating phrases for information retrieval systems. The method does not use an evaluation approach with pre defined queries and calculating recall and precision. Instead, it compares two different techniques by asking people to perform tasks and then looks at the outcomes of those tasks. It is shown that people found phrases automatically generated by finding repeating content words in documents to be good content indi cators and discriminators and were better than using words alone or phrases generated randomly from content words.