home  
  The limit of search engines  
  What are metadata?  
  Subject- and reader oriented metadata  
  How to improve metadata?  
  To what information should we add metadata?  
  Faceted interfaces  
  Where do you store metadata?  
  How to measure reader performance?  
  How to measure information retrieval?  
  How to test information retrieval?  
  Improve the interface  
  Show related information  
  Provide language aids  
  Example  
  Contact Marcel van Mackelenbergh  
  Reactions  
How to test information retrieval?
Goal

The goal of an information retrieval test is to check whether the environment fulfills all the information retrieval requirements.

Overview

For an information retrieval test you need to:

  1. Setup the test
    1. Select the tasks you will test
    2. Select the subjects of experiment
    3. Create a setting that resembles the real environment
  2. Conduct the experiment
    1. Explain to the subject of experiment, what will be tested and how you will use the test results
    2. Give the subject of experiment a short exercise in thinking out loud
    3. Provide the task instruction
    4. Check whether everything is clear
    5. Start the recording
    6. When the reader has come to an endpoint, stop the recording and
    7. Request the reader to fill out a satisfaction questionnaire
  3. Analyze the results
    1. Examine the materials
    2. Notice what behavior occurred almost every time
    3. Write down your impressions
    4. Put your impressions into overviews
    5. Compare the measurements with the information retrieval requirements

Tips

  1. A lot of usability tests have drawn wrong conclusions because they were not testing the reality. Therefore, make sure the environment, the task and the subject of experiment reflect reality. Do not assume.
  2. Prepare. Make sure you have good instructions and explanations. Do a dry run if you can.
  3. The best way of getting results is looking and making notes of everything that strikes you. You can make all kinds of other recordings to: audio, video, log-files However, I find information that comes simply from looking, most valuable.
  4. The only intervention you can make during the experiment is to ask the subject of experiment what (s)he is thinking ("keep on thinking out loud, please"). You should not make any other remarks because they will influence the thinking of the subject.
  5. Keep an open look before focussing on the information retrieval requirements