| Spell checking and syntax checking are two major tasks of text processing. Research on computer techniques for spell and syntax checking has been in progress since the 1960s, but the techniques developed so far still leave room for improvements. This dissertation presents new techniques of processing English text intelligently that are different from the commonly used approaches for text processing.; We consider the user's spelling and typing behavior when detecting and correcting errors. This new approach improves performance by intelligently adapting spell checking to an individual user. Instead of parsing a sentence, we carry out the syntax checking task with new strategies and algorithms. Learning and reasoning are two key features of the methods we propose. Learning during spell and syntax checking involves on-line learning of the user's spelling, typing, and writing behavior. Reasoning techniques are employed to diagnose errors, suggest candidate words, and analyze sentences.; Based on the text processing techniques discussed in the dissertation, we have designed and implemented a new system for processing English text intelligently, called the Laempel System. The system is coded in C under the Unix environment. The reasoning of the system is formulated and carried out in propositional logic, using the Leibniz System for logic programming. The Laempel System has been released. It is being used by several universities and research institutes. The system is also being used as a building block in other research projects.; This research was supported in part by the Office of Naval Research under Grant N00014-93-1-0096. |
