An analysis of user input to an X.500 white pages directory service

This paper analyzes user input to an X.500-based white pages directory service. The motivation for this study is that we can only optimize querying algorithms if we know what sort of queries people make. For example, do people mostly enter forenames and surnames, or surnames only; do people usually enter long organization names or sets of initials; do users enter the two-letter country codes. An area of key interest is how closely the names in users' queries correspond to directory names, as this helps us to determine the types of operation we need to resolve queries. The analysis is both for typed input, where the user is prompted for a set of name components, and also for input in the user-friendly name (UFN) format. We see that the free format of UFN's makes it much harder for users to frame queries which are likely to be successful. We see that while there is poor correspondence between user input and directory distinguished names, much of the user input can be matched exactly with names in the directory, due to the use of alternative names. A number of suggestions are made for transforming user input to increase the chances of finding the target entries.<
>