Interaction between computers and their users is not always straightforward. Potential difficulties and problems in dealing with computer systems could be avoided employing effective measures of assistance.

When users need help for software applications, they usually use internet search engines to get the information they need. However this approach reveals some drawbacks:

• Especially users new to a program often have difficulties specifying appropriate search queries as they are not yet familiar with the "jargon" used for certain software products and thus do not use suitable keywords
• Especially for new software or after updates, it takes some time till relevant information is available in newsgroups or blogs
• Information found often refers to other versions of the software and might be useless for the user's version

Those problems do not apply for help systems that are integrated into the actual software program. While many programs are shipped with an integrated help system users tend to ignore this information source.

One reason for this neglect of integrated help systems might be the fact that those have changed little over time and usually consist of a number of indexed and hyperlinked text files that are arranged in sections and subsections and provide some
kind of search capability.

One of the most famous attempts to introduce a new system for user assistance was surely “Clippy the paper clip”, an interactive animated character that was part of the Microsoft office suite. However, Clippy was widely reviewed negatively by users.

To find out about the origins for the resentments against Clippy and identify solutions and workarounds that make interactive assistants more appealing to users, is the key objective of my thesis. Therefore an interactive, anthropomorphic user interface agent has to be implemented and user studies have to be performed using this agent.