Graphics Reference
In-Depth Information
Figure 4. The phase tagging deque, based on FSMs and HRG structures.
These processes are implemented by using FSMs and common HRG
structures (right-hand side in Figure 4). Each word within the HRG
structure is stored in a Word relation layer. The deque processes POS-
tagged word sequences, and matched them against semiotic grammar 's
rules and relations stored in the off-line constructed gestural dictionary.
Several FSMs are used in order to identify the meaning, and to select
the physical representation of the meaning. Furthermore, FSMs also
predict how the propagation of meaning could be performed over the
specific text sequence (particularly during the preparation movement-
phase, and within the concepts of repetitive, circular, enumerative
gestures). In this way, the deque performs synchronization of the
form. The process of synchronization of the form proceeds as follows.
Firstly, the semiotic tagging process tags those words that could trigger
coverbal movements (mandatory word-type-units). At the morphology
level (based on semiotic grammar 's rules), the algorithm searches for
nouns, verbs, pronouns, fillers and determiners.
Next, the mandatory word-type-units and their neighbors are
grouped into word phrases, and compared against semiotic word-
patterns, in order to define the content and suggest the structure of
the coverbal expressions. As seen in Figure 4 (right-hand-side), the
semiotic processing is traversing through the HRG's Word relation
layer, and by accessing preceding words (e.g. p.pos, pp.pos, ppp.
pos), and succeeding words (e.g. n.pos, nn.pos, nnn.pos) detect
those patterns that match one (or more) of the word-type-orders
defined within semiotic rules. In this way, e.g. the noun represents
the mandatory word-type-unit that indicates existence of coverbal
movement according to the semiotic grammar. By combining the noun
POS tag with its predecessors' POS tags (in this case the adjective
Search WWH ::




Custom Search