T-rule

Revision as of 15:45, 31 May 2013

T-rules, or transformation rules, are rules that alter the state of the nodes. They are used for normalization, for syntactic analysis and for semantic interpretation. The set of the t-rules form the Transformation grammar, or T-Grammar.

Syntax

As describe in the UNL Grammar Specs, T-rules follow the general formalism:

<INITIAL STATE>:=<FINAL STATE>;

And may be classified according to the type of modification that they promote:

• LL, or list-to-list, where the initial state and the final state are list structures
• TT, or tree-to-tree, where the initial state and the final state are tree structures
• NN, or network-to-network, where the initial state and the final state are network structures
• LT, or list-to-tree, converts lists into trees
• TL, or tree-to-list, converts trees into lists
• TN, or tree-to-networks, converts trees into networks
• NT, or network-to-trees, converts networks into trees

Examples

LL rules

• (BLK):=; deletes a node that has the feature BLK
• (A):=(+B); add the feature B to a node having the feature A
• (A):=(-A); remove the feature A from a node having the feature A
• (A):=(B)(C); replaces a node having the feature A by two nodes having the features B and C, respectively
• (A)(B):=(C); replaces two nodes having the features A and B, respectively, by a node having the feature C

TT rules

• XB(A;B):=; deletes the relation XB between two nodes having the features A e B, respectively]
• XB(A;B):=XA(C;D); replaces the relation XB between two nodes having the features A and B, respectively, by a a relation XA between two nodes having the features C and D
• XB(A;B):=XB(+C;-B); adds the feature C to the source argument and removes the feature B from the target argument of a relation XB where the the source node has the feature A and the target node has the feature B
• XB(A;B):=+XC(C;D); adds the relation XC between two nodes having the features C and D, respectively, to the graph where exists a relation XB between two nodes having the feature A and B, respectively

NN rules

• agt(A;B):=; deletes the relation agt between two nodes having the features A e B, respectively]
• agt(A;B):=obj(C;D); replaces the relation agt between two nodes having the features A and B, respectively, by a a relation obj between two nodes having the features C and D
• agt(A;B):=agt(+C;-B); adds the feature C to the source argument and removes the feature B from the target argument of a relation agt where the the source node has the feature A and the target node has the feature B

LT rules

• (A)(B):=XB(C;D); replaces two nodes having the features A and B respectively by a relation XB between two nodes having the features C and D, respectively

TL rules

• XB(C;D):=(A)(B); replaces the relation XB between two nodes having the features C and D, respectively, by two nodes having the features A and B, respectively

TN rules

• XB(%a;%b):=agt(%a;%b); replaces the relation XB between two nodes %a and %b by a relation agt between the same nodes

NT rules

• agt(%a;%b):=XB(%a;%b); replaces the relation agt between two nodes %a and %b by a relation XB between the same nodes

Special types of T-rules

According to their behavior, T-rules may also be classified in:

• A-rules (affixation rules) apply over isolated word forms (as to generate possible inflections);
• L-rules (linear rules) apply over lists of word forms (as to provide transformations in the surface structure);
• S-rules (syntactic rules) apply over trees (as to modify the syntactic configuration).

Further information

For further information on T-rules, see the UNL Grammar Specs