Tuesday, October 14, 2014

A software facilitates computer programming language … – Trends 21


The computers have their own language, so they can only be programmed by those who know the code. A team of computer scientists in the (KIT), Germany, is working on software to facilitate human-machine communication, directly translating the natural language understandable by the computer codes.

As a result, any user could create their own applications with a few sentences. The difficulty lies in the complexity of natural language, do not always processes are described in a strictly chronological order. That is precisely the focus of research in the KIT, for what you have created a tool to automatically sort commands how the machine should run.

“We want to flee complicated rules for users and go to a smart computers to engage in dialogue with us,” explains Mathias Landhäußer scientist Institute for Program Structures and Data Organization (ITP) in the KIT statement released by the university. So far, a program can only be controlled through language if it is designed specifically for this purpose by the manufacturer. An example is sending text messages from a smartphone.

KIT researchers beyond, working on software that you install a language interface for any program. In this way, users will not only be authorized to open, but to manage their applications using voice commands. Currently, the interface has been successfully incorporated into an application that controls the heating, lighting and windows smart homes.


The problem of order

But not everything is rosy. “It will take some time not only that such complex software is operational, but also in natural language to program,” says Landhäußer. Until now, the main obstacle of communication between man and machine, the problem of order, has been solved using English as reference.

Landhäußer exemplified the phrase “Before starting the car, the garage door opens,” a “quite common in our language” description highlights. However, if the process is to running in the virtual world on a computer, a problem arises because the team executes commands successively in the order of arrival.

Returning to the same example, the computer receives the information first “car boot” and then “garage door open”. Therefore, the car crash into the garage door. “If a chain of actions is not foreseen in the program, in the best case nothing happens. At worst, the computer freezes “explains computer.

In contrast, the new software developed at KIT analyzes what was called warning words, usually adverbs or time dependent indicating that the spoken text does not describe the order of the process in a strictly linear fashion. These words of warning clarify if something is done “before” or “after”, “first” or “last”, regardless of the position information in a sentence.

Researchers logical formulas assigned to these words, in order to generate a chronological order in the source text. Applied to the above example, the formula for the signal word “before” the main clause moves automatically to the front. So the result would be “a garage door opens before starting the car.”


Adaptation Machine

The study bet this line after ruling out the possibility of an oral communication adapted to the computer as a reliable alternative. Early tests show that both people with no programming skills as connoisseurs are unable to speak in a strictly chronological order, even if asked to do so. Instead, they are unconsciously using words of warning. “Our goal is that your computer meets the user’s speech, and not vice versa,” qualifies Landhäußer.

In addition to the problem of order, researchers have identified other challenges in natural language programming. Individuals who underwent the experiment some words replaced by synonyms or pronouns. Computers do not automatically understand the term “car” means the same as “carrier” or “that” in the next sentence.

“People understand these relationships because the situation is projected as a movie on your retina. We are working to give computers the ability to understand the long term “, explains the scientist.

LikeTweet

No comments:

Post a Comment