Why Go Physical Linguistics?
From mathematical point of view, Physical Linguistics can not avoid the fundamental
mathematical difficulties as for all other information processing platforms. Therefore,
at the first sight, many people might think Physical Linguistics is a yet-another-paper-maker
kind of topic. However, the following advantages of Physical Linguistics make it outperform
many other information processing platforms just like human natural languages outperform many
coding schemes.
Very High Reusability. While it might be very difficult to reuse a piece a program coded by the former employees in your company, you can reuse all technical documents very easily. Why? Because that piece of program was written in formal language while those technical documents were written in your natural language. Let me show you some examples.
This is the program that is difficult to reuse.
If( input(0.19780405)== "kykyky" )
{
a = 3;
b = 7;
}
This is because you don't know what the string “0.19780405” means and what is ”kykyky” and to make you mad, what the heck do “a” and “b” stand for? And why they must be equal to 3 and 7, respectively? Well, when you read the technical document, everything should be so simple to understand, for example:
“If on April 5, 1978 we got an input from Tokoyo, then we price water to $3/per gal and gas $7/per gal.”
If oneday your manager suddenly tells you: We need to change our model into
“If on May 4, 1990 we got an input from Shanghai, then we price water to $5/per gal and gas $8/per gal.”
Please finish it within 30 seconds!
Which way will you choose to revise? A piece of program or a simple sentence? This is exactly the
reason why the engineers in YangSky feel very happy to use physical linguistics.
High Prototype Speed. Try to build a prototype of a new product sometimes takes huge amount of human-years with first-class well-trained engineers and scientists. For example, if we want to control an electrical machine to a very high accuracy, we need the best electrical engineers to derive the detailed model of that electrical motor and then need the first-class engineers to design the controllers. OK, suppose we just finished our design and suddenly the head of the sales department tells us that the market needs another kind of electrical machine to be controlled. What should we do? We need to do it all over again! To find the detailed model, to calculate all parameters for the controller and implement it!
In YangSky, our engineers know physical linguistics and didn't need to do in this way.
First, instead of a detailed mathematical model of the electrical motor with many equations,
a semantic model for the electrical motor was first built up. Then for each component in this
semantic model, the computational verbs and nouns are modeled. Then based on this model,
computational verb fuzzy controllers are built. Once these step finished, the knowledge
of the control subject and the controller was put into modules. If one wants to change the
type of the motor, it is perfectly fine because we can just simply change the corresponding
computational verb rules without destroy the wholeness of the entire system. Therefore, all
prototypes along a product line can share the same physical linguistic model with locally
modified parameters or rules.
The tricks the engineers in YangSky used are always to keep the knowledge encoded in natural
languages based on physical linguistics and to make complicated tasks into decouples
sub-modules. The risk of the entire project can be controlled at each stage and
the knowledge can be reused over the entire institute and inter-generation of engineers.
See PL Image Understanding Engine to believe.