Example of a Task out of Lab Station 5:
Semantics of prepositional phrases
Upgrade the existing parser to be able to understand all prepositional
phrases built with "on
Prepositional phrases (PPs) are constituents of sentences introduced by
a preposition like "on", "over", "in", and so on. The automatic
interpretation of PPs is very important for understanding and
disambiguating natural language sentences. Computational linguists use
special programs, so-called parsers, to analyze natural language
expressions. Parsers must "know" the rules, according to which they can
analyze the language. This set of rules is called a grammar. Every
grammar for a natural language must also have a subset of rules that
cover prepositional phrases. These PP-rules can be formulated with the
expressional means of MultiNet
The description of this meaning representation language can be found in
the MultiNet documentation (H. Helbig, 2005).
- Helbig, H., (2005). Knowledge
Representation and the Semantics of Natural Language Springer,
The chapter about relations and functions is of special help. Not all
mentioned relations are relevant for this task; important for solving
this task are the following (and maybe some associated) relations:
ASSOC, ATTR, CAUS, CIRC, DIRCL, DUR, INSTR, LOC, OBJ, ORNT, RSLT, TEMP,
VAL, *AUF (as subfunction of *FLP).
- Helbig, H. and Gnörlich, C. (2002). Multilayered
Networks as a Language for Meaning Representation in NLP Systems.
In: Lecture Notes in Computer Science 2275, Springer, Berlin, p. 69-85
This paper is
a summary of MultiNet.
- Hartrumpf, S. (1999). Hybrid
disambiguation of prepositional phrase attachment and interpretation.
In Proceedings of the Joint Conference on Empirical Methods in Natural
Language Processing and Very Large Corpora (EMNLP/VLC-99), p. 111-120
In this paper, the use of PP-rules in a complete module for
disambiguation is described.
- Helbig, H. and Hartrumpf, S. (1997). Word
class functions for syntactic-semantic analysis . In Proceedings
of the 2nd International Conference on Recent Advances in Natural
Language Processing (RANLP'97), S. 312-317. Tzigov Chark, Bulgaria
The parser is documented in this paper.
The rest of the task is left out.