What is transaction log analysis?

Transaction log analysis is a way to track unobtrusively how users are using a digital library. As an evaluator, you may wish to analyze transaction log information as part of an overall evaluation aimed at obtaining a deeper understanding of how users are navigating through your digital library, which resources they access, and any search problems they encounter. Log analysis alone usually requires too much inference, but it provides important information that may be explained with the additional data obtained from interviews, surveys, and observations. By understanding what digital library users are doing within your digital library, you can then make informed decisions about how to better meet users' needs, perhaps by improving the quality of the underlying search algorithms or enhancing the graphical user interface.

Transaction logs are usually gathered through transaction monitoring software that is typically built into a digital library system or based on a web server that automatically tracks specific interactions. Most digital libraries maintain server logs that keep track of users' requests. These log files typically contain information such as the users' IP addresses, date and time of users' requests, search terms, and so forth. As an evaluator, analyzing user logs can provide valuable information, e.g., it can be used to create a map of what a typical user session looks like.

Here are some of the types of information that transaction log analysis can provide for evaluating digital libraries (Tenopir, 2003):

• Frequency of feature use
• Sequence of feature use
• System response times
• Hit rates
• Error rates
• User actions to recover from errors
• Number of simultaneous users
• User session lengths
• Number of transactions per session
• Location of users