Date of Defense
A natural language is a language which is used by humans, having evolved through the thousands of years of human existence. To understand natural language processing, one needs to understand the definition of words such as diction, grammar, parsing, semantics, and syntax. One possible definition for natural language processing is the use of syntax in a grammar by a parser to understand a natural language. After discussing the purpose and possible uses for natural language processing, this paper discusses many different systems, some of which make use of variations on this definition, and some of which use more advanced ideas. It also talks about problems encountered in natural language processing, and then goes on to discuss in brief some theories in natural language processing.
Natural language processing is the subject of much artificial intelligence research because of the many possible uses for it. For example, a natural language processor could be used for increased user-friendliness, such as with natural language database front ends. Increased understanding of human intelligence is also possible through natural language processing research.
There are many systems which have been written as natural language processors. They range from the early key word systems to the later semantic analyzers. Many systems have been written as database front ends, such as LUNAR or PLANES. Other systems, such as FRUMP, IPP, and McMAP, implement what is called realistic language comprehension to process newspaper stories. Other implementations include PAM, SAM, and ABDUL/ILANA.
Problems are often encountered in writing natural language processing systems. The problems which one runs into in writing a natural language processor include word ambiguity, ambiguous references, idioms, ellipsis, and verb tense. Although the solutions to these problems often seem simple in theory, the implementations are what lead to the many different theories of natural language processing.
Some theories in natural language processing include syntactic analysis, scripts, augmented transition networks, and conceptual dependency. The various merits of these methods are discussed.
Turner, Stephen W., "A Research Report on Natural Language Processing" (1986). Honors Theses. 1500.
Honors Thesis-Campus Only