The life of a software creator consists mostly in staying in chair in front of the computer. As I’m a passionate gamer too, this started to be an issue as started to feel some pain in back. That’s why I’ve decided to take an action and chose the best gaming chair to have a better support for my back while coding or gaming.
A word lattice is a directed acyclic graph with a single start point and edges labeled with a word and weight. Unlike confusion networks which additionally impose the requirement that every path must pass through every node, word lattices can represent any finite set of strings. In general a word lattice can represent an exponential number of sentences in polynomial space. Here is an example lattice showing possible ways of splitting some compound words in German:
Representing lattice input
Lattices are encoded by ordering the nodes in a topological ordering and using this ordering to assign consecutive numerical IDs to the nodes. Then, proceeding in order through the nodes, each node lists its outgoing edges and the cost associated with them. For example, the above lattice can be written as:
( ( ('einen', 0, 1) ), ( ('wettbewerbsbedingten', 0, 2), ('wettbewerbs', -0.2, 1), ('wettbewerb', -1.5, 1) ), ( ('bedingten', 0, 1) ), ( ('preissturz', 0, 2), ('preis', -2.0, 1) ), ( ('sturz', -1.0, 1) ) )
For each edge, the first field is a string representing the edge label, followed by a cost and the distance between the start and the end node of the edge. The string
'*EPS*' indicates an espilon edge. This format is informally called the “Python lattice format” (PLF).
cdec accepts PLF lattices written on a single line as input. The edge cost value is associated with a feature weight named
For more details on lattice translation, refer to:
Chris Dyer, Smaranda Muresan, and Philip Resnik. Generalizing Word Lattice Translation. In Proceedings of ACL 2008.