Different Keystrokes for Different Folks: Visualizing Crowdworker Querying Behavior


Search engine users retrieve relevant information for an information need using keyword queries. Different users may have similar information needs, but use different query terms. The resulting user query variations can provide a wealth of useful information to IR researchers. Most recently, the keystroke-level telemetry data gathered as part of the CC-News-En collection provides important insights into how users create queries for a search task, at a level of detail not possible using a normal query log. In this demo, we present an interactive tool that enables practitioners to visualize users formulating queries. Our new tool is a temporal simulation of the typing behavior of crowdworkers, grouped by information need. It provides the ability to directly compare the cognitive behavior of multiple users simultaneously, and observe how query keyword selection and ordering happens before a final query is submitted to a search engine. To demonstrate the benefit of our tool, we include a qualitative study of four different user behavior patterns which were observed in the CC-News-En collection.

Proceedings of the 6th ACM SIGIR Conference on Human Information Interaction and Retrieval (CHIIR 2021)