Game Development Reference
In-Depth Information
1. When two or more players agree on the same tag over a single image, that tag
is declared valid. This is the original cross-player scheme. When the pool of
players is very narrow, it becomes very restrictive (although it is precise). Yet with
the additional player motivation and purpose awareness, it might be considered
unnecessarily strong, so we introduced several “weaker” ones.
2. When the same player repeats a certain tag for a given image (which may happen
if the image is used in more than one game), it is declared valid. The repetition
of terms (even by a single player) may indicate that their use was not a mistake.
3. When the same tag was used (sufficient number of times in total) by one or
more players to describe various images within the same album, all of these tag
assignments are declared valid (the “sufficient” value used in our experiments
was four). This rule takes advantage of the possible recurrence of same concepts
within a single image album, such as person names.
5.4 PexAce Personal: Evaluation
We have performed a combined quantitative-qualitative study to evaluate the PexAce-
Personal. The study included the gameplay itself, so no prior game deployment was
performed. We examined:
1. Validity of the image tags acquired through game.
2. To what extent are these tags “image-owner-specific”.
3. The types of tags extracted as well as performance of the game over different
types of images.
4. How the game performs, when players are unaware of its purpose (with respect
to the possible scenario of sharing the game within one's social circle).
Research questions.
We have defined following research questions, under circum-
stances that the game is played with low number of players with their personal
images.
1. Do tags describe the images correctly (measure of correctness (precision)—a ratio
of correctly assigned tags to all assigned tags)?
2. How specific (to the social group, to which the images belong) are the correctly
assigned tags (specificity)? How are the tags understandable (to persons outside
the social group) (understandability)? The understandability (for “outsiders”) is
a minor factor for us, yet it says something about the usefulness of the tags. We
expected the understandability to be inversely correlating with specificity.
3. What effect does the awareness about the game's purpose have on the measured
characteristics and the perception of the game by players?
4. In what quantities are the following metadata types—person names, places and
events—present in the acquired tag sets?
5. For which type of images is the approach suitable. We have defined several image
categories orthogonal to their topics: portraits, groups, situational (may contain
persons, but the dominant feature is some situation) and no-persons images.
Search WWH ::
Custom Search