曝光台 注意防骗
网曝天猫店富美金盛家居专营店坑蒙拐骗欺诈消费者
or more passengers.
In 1995, SMART Communications, Inc. (SMART) was commissioned by Nortel Networks, a
large global manufacturer of telecommunications equipment, to develop a controlled language. At
that time, the global demand for telephone switches expanded rapidly for the Internet. Nortel did
not have enough technical translators to keep up with the demand for multilingual documentation.
SMART used the steps outlined in this paper to develop Nortel Standard English. All of the
technical terms were developed with the help of the SMART Text Miner. The subject matter
analysis was done by Dr. Ralph Calistro, of Nortel Customer documentation staff, in Ottawa,
Canada. During the project, SMART and Nortel saw the need for a sophisticated grammar checker
to help their busy writers. To meet this need, SMART wrote the MAXit Checker that uses
artificial intelligence to read, understand and electronically critique Controlled English sentences.
The MAXit Checker now has a decade of use by thousands of writers around the world. The
MAXit Checker is a plug-in to Adobe FrameMaker, WORD, Arbortext Editor and other tools.
The writers at Nortel write their documentation in structured (Extensible Markup Language)
XML, then ask the MAXit Checker to check for compliance. This inexpensive, quality-control
technique has eliminated some 42% of procedural errors and
made complex telecommunications documentation easier to read and understand in 208 countries.
There is an example shown later in this paper.
How to Develop a Controlled English Dictionary
There are three easy steps, which are also shown in the demonstration:
Step 1: Text Mining: This task is the extraction of knowledge from existing documents by a
computer. The documents are processed by the SMART Text Miner to generate a report, as
shown below. This report allows the user to quickly sort the good terms from the bad terms and
store that knowledge in a database. These databases are distributed on a server to all writers and
authors.
In the picture above, from left to right, column one holds an index to the original document.
Column two shows the left context. Column three shows the key word. Column four shows the
right context. Column five shows the number of occurrences the key word is found in a set of
documents. The higher the frequency of occurrence of word in the text, the more important the
term.
This software is easy-to-use by persons who are not lexicographers or linguists.
The terms are extracted and checked by subject matter experts. In this example, the user can
quickly identify a “positive displacement pump”, which is a valid term with a valid NATO stock
number. Another term that is easy to identify is “potassium chloride”, a valid chemical name.
The Text Miner also identifies ambiguous terms like “potential”, which can mean a measure of
electrical force or an adjective. These terms are put in a reverse dictionary for use in the MAXit
Checker. The Text Miner has a convenient browse function, term harvester and Mining-at-glance
feature. The objective is to discover the hidden terminology that must be included in a Controlled
English dictionary. It is possible to develop dictionaries manually, but the computer-approach is
fast, accurate and more efficient.
Step 2: Electronic Critique – MAXit Checker
It is possible to manually check texts for compliance to a Controlled English vocabulary, but the
manual method is too slow. The key to a successful Controlled English program is to give the
technical writers and authors a tool that does 95% of the work, but leaves the final decision to
writer.
The MAXit Checker is such a tool and integrated in major publishing tools to extend the grammar
checking functions. Because MAXit uses artifical intelligence, the writer has the illusion that
computer is thinking. The MAXit Checker includes some 9,500 grammar rules, based on style
guides from public sources and the Chicago Manual of Style. Because the MAXit Checker uses
customized dictionaries with linguistic tags, the software can make sophisticated checks for terms
like ExxonMobil and mHz.
The MAXit Checker can also disambiguate terms like db for decibels and database.
A metric conversion function helps authors express dual dimensions.
The next page is a picture of the MAXit Checker inside a WORD document. The colors, added
by the MAXit Checker, indicate the possible errors in 40 message classes.
For example, the words marked in yellow must be examined for possible addition to the technical
dictionaries. A like term “peak efficiency” is both difficult to define and translate. The writer is
prompted for a definition, perhaps revolutions per minute, number of operating hours or mean time
中国航空网 www.aero.cn
航空翻译 www.aviation.cn
本文链接地址：
航空资料28(62)
