State-of-the-Art: Design of the Semantics Acquisition Games - Semantic Acquisition Games: Harnessing Manpower for Creating Semantics - page 93

which emerged in the last decade as web browser games, are gaining more and more
players different from players of classical video games: while the typical player
of classical video game is a teenage male playing for extensive amount of time a
day, the casual games players are of all age categories and of both genders, playing
significantly lesser. The potential SAG players recruit from this group, which means
that SAGs do not have to compete with elaborated classical video games (which are
costly results of years of development), they could be small applications with simple
game rules and short play time, to “fill a coffee break time of an mid-aged female
office employee”.
7.1.2 Von Ahn's Approach
An attempt to formulate a SAG design methodology has been made by von Ahn
and Dabbish [ 28 ]. Their aim was to come up with an universal abstract scheme
for resource description acquisition game, use of which would only depend on the
actual resource type. In the game, the player is confronted with a resource and then
somehow tricked into directly or indirectly describing it. The authors suggested that
a “game with a purpose” should follow one of the three abstract schemes. All of
them are two-player schemes (so the players can mutually validate their actions, i.e.
the metadata they produce), and use the effect of unconventional socialization and
hidden information about the game state (in all schemes, players cooperate, but are
unable to communicate with each other).
1. Output agreement scheme . In this scheme, the players are given the same
resource and then try to blindly agree on a word that describes it. A principle
typical for the ESP game [ 27 ].
2. Input agreement scheme . In this scheme, the players are presented with
a resource sample separately. It is either the same for both players or differ-
ent for each player. After the players see their resource, they describe them and
according to this descriptions, the players decide whether they see (hear) the same
thing or not. A typical example of this principle is the TagATune game [ 12 ].
3. Inverse problem scheme . In this scheme, the players have not the same role,
instead one is confronted with a resource while other has to guess or correctly
select it according to clues created by the first player (which are later mined
for metadata). A typical SAG using this scheme is Peekaboom [ 29 ]—there the
resource is a term which the other player is guessing.
Ahn's ideas have been implemented in many SAGs (his own, but also from other
authors). The mechanisms he designed cover a fairly broad group of game pur-
poses in semantics acquisition—mainly involving (multimedia) resource descrip-
tions acquisition. On the other hand, the proposed mechanisms can hardly be used
everywhere (e.g. ontology linking, co-reference identification or outside the seman-
tics acquisition, e.g. protein folding or FPGA optimization) and also impose several