NLP DATABASE

Nlp DB

Database: How

Noun (table)

Column name (field)

Location of decision tree

Column name (field)

Location of decision tree

The question will be parsed to obtain the best guess at the parts of speech present. These parts of speech will be ran through several filters to generate a SQL

query that will then be used to retrieve the proper decision tree.

NOTES:

• Please realize this is a prototype and future work will be done not only by me but hopefully others there are many things that will change neither I or anyone else solved the problems with natural language processing and it is doubtful that it will be “perfected” anytime soon.

• Using a more efficient trainable parser would greatly improve accuracy rate.

• At this time there are separate databases pertaining to each questions How, What, When, Where, Why this will result in too much redundancy a total re-working will be forthcoming with more time put into the design. There are much more efficient ways to access this data.

• This is still in the very early stages. I realize this solution will probably be inadequate for the task. The only advantage I can see is to rely on the Decision trees to do the work thus limiting how close my “best guess” has to be.

• Going to include a synonym dictionary to help narrow down the correct noun and reduce redundant information being stored.

• Anything including a number or a measurement will be much easier to handle. The primary noun will usually follow directly afterwards.

• Primary task of this coming semester will be researching methods to improve the accuracy rate of answering the question correctly.