| |
|
|
|
|
|
|
|
|
| |
|
|
|
|
|
|
|
|
|
| Last modified:
2006-11-08 |
|
The DFKI Evaluation Center for Language Technology |
| |
Recent
Work
|
|
| |
Automatic Hyperlinking |
|
| |
- Questions:
- How should hyperlinks be marked in text?
- Trade-off between distraction and link awareness
- Increased relevance of link redundancy
- Which occurrences of a term should be hyperlinked?
- All occurrences – too distracting?
- Linguistic selection criteria?
- Which occurrences do users actually use?
- Trade-off between distraction and accessibility of linked information
- Results:
- Users tend to use one of the initial occurrences of a link even
if they do not read the text linearly
- Problems caused by adjacent full form and abbreviation of same concept
- E.g. „Gesellschaft bürgerlichen Rechts (GbR)“
- Link intuitiveness – cognitive overhead
- Users do fix hyperlinks they do not follow
- Some users still complained links hard to find
- Repetition of information in text and hyperlinks viewed as negative
- Users were annoyed by hyperlinks containing little information
- Study performed with the ASL Model 504 remote eye tracker
- Screenshot of the text used (click on the picture to enlarge it)
- Fixplots (click on the picture to enlarge
it)
The red lines correspond to the user's line of gaze, the
red dots to the fixations: the longer the fixation, the
bigger the dot. The fixations are numbered (number next
to the fixation) and start with "1".
|
|
|
