A search engine log analysis of music-related web searching

Abstract

We explore music search behavior by identifying music-related queries in a large (over 20 million queries) search engine log, gathered over three months in 2006. Music searching is a significant information behavior: approximately 15% of users conduct at least one music search in the time period studied, and approximately 1.35% of search activities are connected to music. We describe the structural characteristics of music searches—query length and frequency for result selection—and also summarize the most frequently occurring search terms and destinations. The findings are compared to earlier studies of general search engine behavior and to qualitative studies of natural language music information needs statements. The results suggest the need for specialized music search facilities and provide implications for the design of a music information retrieval system.