5 - Summary

In chapter 1 - Introduction we defined several important terms; among these were information, communication and language. All communication is driven by the need for exchange of knowledge and is based on common protocols like signs and grammars.

These can be realized in computer applications, as could be seen in chapter 2 - Technology Applied. Several implementations were presented which all deal with a specific aspect of natural language processing.
Translation was dealt with in an example of German-English machine translation which also permitted us to see the restrictions of current implementations; these are a limited world knowledge base and therefore a limited domain of discourse, and there is the concept of restricted languages - languages that are simple enough to be understood by automatic translators.
The Database Access paragraph addressed the problems connected with accessing databases with natural language input - these difficulties are similar to the ones that translator applications face.
Dealing with text included issues like full-text search algorithms and methods (information retrieval), text categorization (sorting texts according to their topics), and data extraction.

In chapter 3 - Syntax, the basics of formal languages were introduced: (non-)terminal symbols, rewrite rules, phrases, lexicons, and several classes of grammars that describe their generative capacity.
Parsing and Efficiency and the next two chapters explained the task of parsers (disassembling the input text back into nonterminal symbols and phrases) and devised two different parsers, one that works straightforward and is easily outperformed by the second parser (the chart parser) in terms of efficiency and error tolerance. The process of tokenization and analysis of distinct words was addressed, too.

In 4 - Semantics, we showed how the inclusion of meaning into the parsing process can affect the result of the parse. Several semantic relationships were presented as well, together with a formal representation of the semantics of a phrase. Finally, we said that discourse understanding relies heavily on semantic analysis and incorporation of external knowledge.

back to main page...