Abstract: How HCI systems can adapt to the changing retrieval tasks of users. It is a topic to be studied to provide better interaction modalities. Thirty subjects were invited for the experiment. A two-factor between-subjects design of 3 (interaction modality: touch, voice, and eye-control) × 2 (task type: goal-ambiguous task, goal-oriented task) was used for the experiment. Objective data such as task completion time and number of gaze in the area of interest were tested by eye-movement experiment, and subjective data such as task load and satisfaction were tested by questionnaire after the experiment. The experimental results showed that both task type and interaction modality affected user retrieval, and there was an interaction effect. The clearer the user's retrieval goal, the faster the task completion time, the more focused the attention area, the lower the task load, and the higher the satisfaction level. Users retrieved more slowly using eye-control and voice modality than using touch modality. The study provides design reference for the selection and use of interaction modalities in information retrieval systems。