ADAPTIVE INTERFACES AND AGENTS
Anthony Jameson · 2007
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434 Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434 Functions: Supporting System Use . . . . . . . . . . . . . . . 435 Taking Over Parts of Routine Tasks . . . . . . . . . . . . . . . 435 Adapting the Interface . . . . . . . . . . . . . . . . . . . . . . . . . 436 Helping With System Use . . . . . . . . . . . . . . . . . . . . . . . 437 Mediating Interaction with the Real World . . . . . . . . . 438 Controlling a Dialogue . . . . . . . . . . . . . . . . . . . . . . . . . 439 Functions: Supporting Information Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440 Helping Users to Find Information . . . . . . . . . . . . . . . 440 Support for browsing . . . . . . . . . . . . . . . . . . . . . . . 441 Support for query-based search or filtering . . . . . . 442 Spontaneous Provision of Information . . . . . . . . . . . . 442 Recommending Products . . . . . . . . . . . . . . . . . . . . . . . 442 Tailoring Information Presentation . . . . . . . . . . . . . . . 443 Supporting Collaboration . . . . . . . . . . . . . . . . . . . . . . 444 Supporting Learning . . . . . . . . . . . . . . . . . . . . . . . . . . 445 Usability Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . 446 Threats to Predictability and Comprehensibility . . . . . 447 Threats to Controllability . . . . . . . . . . . . . . . . . . . . . . . 447 Obtrusiveness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448 Threats to Privacy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448 Breadth of Experience . . . . . . . . . . . . . . . . . . . . . . . . . 448 Dealing With Trade-offs . . . . . . . . . . . . . . . . . . . . . . . . 448 Obtaining Information About Users . . . . . . . . . . . . . . 449 Explicit Self-Reports and -Assessments . . . . . . . . . . . . . 449 Self-reports about objective personal characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449 Self-assessments of interests and knowledge . . . . 449 Self-reports on specific evaluations . . . . . . . . . . . . 450 Responses to test items . . . . . . . . . . . . . . . . . . . . . 450 Nonexplicit Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450 Naturally Occurring Actions . . . . . . . . . . . . . . . . . . . . . 450 Previously Stored Information . . . . . . . . . . . . . . . . . . . 450 Low-Level Indices of Psychological States . . . . . . . . . . 451 Signals Concerning the Current Surroundings . . . . . . 451 Special Considerations Concerning Empirical Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451 Use of Data Collected With a Nonadaptive System . . . . 451 Early Studies of Usage Scenarios and User Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452 Wizard of Oz Studies . . . . . . . . . . . . . . . . . . . . . . . . . . 452 Comparisons With the Work of Human Designers . . . . 453 Experimental Comparisons of Adaptive and Nonadaptive Systems . . . . . . . . . . . . . . . . . . . . . . . 453 Taking Into Account Individual Differences . . . . . . . . . 454 Checking Usability Under Realistic Conditions . . . . . . 454 The Future of User-Adaptive Systems . . . . . . . . . . . . . 454 Growing Need for User-Adaptivity . . . . . . . . . . . . . . . . 454 Diversity of users and contexts of use . . . . . . . . . . 454 Number and complexity of interactive systems . . . 454 Scope of information to be dealt with . . . . . . . . . . 454 Increasing Feasibility of Successful Adaptation . . . . . . 455 Ways of acquiring information about users . . . . . . 455 Advances in techniques for learning, inference, and decision . . . . . . . . . . . . . . . . . . . . . 455 Attention to empirical methods . . . . . . . . . . . . . . . 455 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455 This chapter covers a broad range of interactive systems. They all have one idea in common: It can be worthwhile for a system to learn something about each individual user and adapt its behavior to them in some nontrivial way.