Soar to success of this project relies on three ideas.
The First is a dictionary in the format of semantically connected associative network. Logical interconnections between words should represent the language structure in interpretable by computer format. For this purpose simple two-dimensional tables are not a good instrument. By two parameters which can be shown in columns and rows we cannot vividly set and analyze the great amount of mutual directed connections. But this network naturally works in our associative brain very well.
Thus, for visualizing semantic network there is a need in special tool: Graph Editor.
The second. Generalizing-classification algorithm
Benefit of such algorithm is considered in consistent patterns discovery of words meaning resemblance. This will provide future system with self-studying ability.
The third. Grammar parser
This will convert information from semantic format to natural language and vice versa.