Authors: Victoria Wang, Kevin Yap
What is the problem?
People make grammatical errors all the time. Our goal is to detect some simple grammatical errors through a rule-based system. We'll need to gather linguistic information from different resources, such as (not limited to):
What is the something extra?
1. Taking punctuation into account.
2. Implementing a bigger and more complicated dictionary & database of facts.
3. Suggesting corrected phrases in addition to simply analyzing sentence structures (like the query system does).
What did we learn from doing this?
In theory, Prolog is suitable for parsing the syntax of a natural language like English, due to its pattern matching capabilities. This is evident from the fact that our project is able to correctly identify proper verb-noun agreement in simple sentences. In practice, it is difficult (impossible?) to compile a suitable dictionary of valid words and their parts of speech. Some sort of predictive engine could be used to guess the type of an "unknown term", but this would be hard to implement in Prolog.