Natural Language Processing : Ultron

GitHub Link

NLP "

Introduction : Text Ontological Relation Extraction

One domain of artificial intelligence is a topic called Natural Language Processing. Fmr an AI to understand and communicate effectively, it has to understand language and semantics of human language. The purpose of the project is to implement a program that is capable building a logical and relationships structure out of sentences, which can be extended to a wall of text.

Topics

The topics covered used in the project includes:

1. Formal Grammar of English
2. Relational Database (SQLite)
3. Parsing with Context-Free Grammars
4. Semantics and Pragmatics

Project Scope

The project is separated into several sub-parts.

1. To extract parts of sentence, a word tokenizer is used.
2. For each word, its attributes can be determined by using a database of dictionary. A public dictionary is used in form of SQLite database. This sub-parts will construct ‘Word’ class and objects that has a token and it’s part of speech objects.
3. From the part of speech objects, a syntax tree is constructed. This syntax tree will be used to identify what kind of sentence given.
4. The last part of the project is to use relational database to build a relationships between part of speech objects.
5. Combining all parts of the projects, the program should be able recognize, remembers and return textual informations given.

Project Diagram

1. Tokenize Input into Tokens
2. Tag Tokens with POS Tags
3. Determine the most probable Tag for each Token
4. Parse Tokens into Syntax Tree using Grammar Structure
5. Determine Head Verb/Noun of the Tree and each of its Sub-Trees
6. Determine the most probable Tree
7. Create Relationship Diagram from the Tree