Modelbased Highlighting to Support Selective Reading on the
Model-based Highlighting to Support Selective Reading on the Web Ion Juvina ijuvina@andrew. cmu. edu
About. . . n n . . . me Information Science @ UU. NL
Aknowledgements n n n n n Herre van Oostendorp Eelco Herder Ellert van den Broek Vincent van der Linden Koen Buurman Poyan Karbor Brian Pauw Arnaud Lek Martijn Abbing Richard van Yperen
Contents n n n n Introduction Study “Individual differences” Cognitive model Study “Voice suggestions” Study “Graphical suggestions” Study “VIP” Conclusion and discussion
Introduction n Navigation is a major part of user experience on the web (Lazar, 2003) n Web interfaces n n Revisits n n Structure vs. content Post valued recall Case: using the web by visually impaired persons (VIP) via screen readers
Study “Individual differences” n Objective: n Identify determinants of performance and satisfaction in web-assisted tasks
Activities n Morrison, Pirolli and Card (2001) Types of activities users perform on the Web. n 2188 usable responses. n Example: I accessed Netscape's financial site to check my credit card balance and how long it would take to pay it off. I'm now MUCH more fiscally aware of my spending habits and am trying to pay off my balance more actively. n
Tasks
Hypothetical factors n n n n Domain expertise Working memory capacity Episodic memory Cognitive style Spatial ability Age Gender
Navigation metrics
Results
Results (cont’d)
Results (cont’d)
Cognitive models n n SNIF-ACT Co. Li. De. S MESA Value and limitations: n Information scent n n n Common Too narrowly defined Semantics prevail over syntax Static background knowledge Common user prevails over individual differences
Co. Li. De. S+
Models vs. data
Study “Voice suggestions”
Study “Graphical suggestions”
Study “Graphical suggestions” (cont’d)
Study “VIP” n Manipulation n Screen switched off Highlighting by increasing reading priority Suggestions generated on-the-fly n n No look ahead Results: n Common mechanism n n Spatial ability – task performance Weak suggestion mechanism n n Position: 5 -> 4 Breaks of coherence
Conclusions n Support for selective reading is: n n Needed Possible Not easy Empirical research and cognitive modeling could help
Discussion points n n n Ecological validity vs. accuracy ? Scalability of cognitive models ? Ontologies vs. semantic spaces ? Goals known in advance ? Static goal representations ?
- Slides: 22