Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 02544238 1996-02-14
iVIETHOD FOR ELECTRONICALLY TRANSLATING TEXT
This application is divided from Canadian Patent Application Serial Number
2,216,426, filed Febniary 14, 1996.
BACKGROUND AND SUMMARY OF THE INVENTION
The present invention relates generally to the technical field of electronic
communication over a computer network and, more particularly, the present
invention
relates to multi-lingual electronic communication in the form of parallel
electronic
discussion groups.
In the field of the present invention there are basically two types of
electronic
network communication in use today: electronic mail and bulletin board
systems.
Electronic mail is a way of sending messages to another computer user
(possibly located
at a remote site) through the use of a modem. Electronic mail is generally
used to send
messages to other computer users about any subject. A bulletin board system
(BBS) may
be run from a computer that uses a special program to allow other computers to
call it, by
use of a modem, ordinarily over standard telephone lines. A BBS acts as a
storage
facility, where people calling from their computers can post or receive
messages and
send or receive program files. A BBS is usually subdivided by topics so that
users with
similar interests can send information to other users of similar interests.
These related
groups of messages are referred to as topics, message areas, Forums (a
registered
trademark of CompuServe Incorporated) or conferences (i.e. a particular BBS
may
contain numerous related groups of messages such as on the subjects of travel,
sports,
stamp collecting, etc...).
BBSs are distinguishable from electronic mail in that BBSs are used for
posting
messages of a particular group of computer users who have similar interests
but these
posted messages are not ordinarily addressed to any particular user. BBSs are
also used
by some computer users who wish to just read posted messages and other
information,
without having to post a reply message. For example, a person in Columbus,
Ohio, who
1
CA 02544238 1996-02-14
is preparing for a vacation, may access a BBS on travel, through his home
computer, and
post a message asking for the best places to go skiing in the United States.
Other
network users with an interest in travel may read the message from the person
from
Columbus, and then reply with their opinion about where the best place is for
skiing.
There are thousands of BBSs available to computer network users on a wide
variety of
subject areas.
As the number of computer networks being accessed by international users
increases, the number of online discussion groups for users with similar
interests, from
different countries, also increases. For example, a person in France may
access a
discussion group on international law, to obtain information about trade laws.
Assuming
that the French person is only familiar with the French language, the French
person
cannot send a message that would be understood by a German speaking user or an
English speaking user. Thus, the need arises for an electronic discussion
group which
can be utilized by foreign language speaking users.
Language translation software is currently available, however, to Applicants'
knowledge, it has only been used for purposes of electronic mail in online
services
environments, where the identity and language of the person receiving the
message is
known. For example, if a French person wants to send a message to Bob Smith,
who is
American, the French person can send the message through a French to English
machine
translator on the network. However, there is currently no successful
implementation of
language translation software in an environment such as an electronic public
discussion
group which may have an unknown group of users of different languages. For
example,
if a French speaking person wants to access a travel discussion group, run
from a BBS in
New York, and inquire into the best places for mountain climbing in the world;
currently, the French person can translate his message to the discussion group
in only
one other language (e.g. English). Thus, only people who can read English or
French
would understand and be able to reply to the message. A person in Japan, who
does not
have command of the English language and has valuable information relating to
mountain
climbing in Japan, will not understand the message posted in English and will
not be able to
2
CA 02544238 1996-02-14
respond to the inquiry. Additionally, if the French person camzot read
English, he will not
understand replies posted to the discussion group in English.
The present invention is a method for electronically translating text in a
source
language. This method finds particular applications in a system of parallel
discussion
groups operated in conjunction with a message collection/posting software
program, and a
data filter software program in which there are a number of discussion groups
running in
"parallel"; one group for each language being used in the discussion groups.
The individual
discussion groups all contain the same information, in the same order; the
only difference
being that each parallel discussion group is written in a different language.
Accordingly, the present invention provides a method for electronically
translating
text in a source language, comprising the steps of: pre-processing said text
with a
software filter to identify a text string within said text to remain
untranslated; placing a
machine readable mark in association with said text string; submitting said
text to an
electronic and automatic language translator, whereby said language translator
translates
said text into a target language in accordance with a machine translation
dictionary while
leaving said text string untranslated; and receiving said translated text and
said
untranslated text string, electronically and automatically from the language
translator.
In a further aspect, the present invention provides a method for
electronically
translating text, comprising the steps of: identifying a text string to remain
untranslated,
said text string included within a text message from an electronic discussion
group;
adding a machine readable mark to said text string to remain untranslated;
submitting
said text message to an electronic, automatic language translator, whereby
said language
translator translates said text message into a different language while
leaving said
marked text string untranslated; receiving translated text message from the
language
translator; and electronically and automatically deleting said mark from said
text string
within said translated text message.
The present invention also provides a method for electronically translating a
text
message in a source language posted to a first electronic discussion group,
comprising
the steps of: identifying within said text message at least one text string to
leave
3
CA 02544238 1996-02-14
untranslated; inserting in said text string a marker to identify said at least
one text string
to leave untranslated; submitting said text message with said marker to an
automated
language translator for translation of said text message from said source
language to a
target language, wherein said automated language translator leaves said at
least one text
string with said marker translated; deleting said marker from said at least
one text string;
and posting said translated text message in said target language to a second
electronic
discussion group.
Once user logs onto a particular parallel discussion group he or she may then
choose his or her language preference. If the user's language reference is set
to French,
the French version of the discussion group will be accessed. Messages posted
to a
discussion group will be periodically collected, translated to the other
languages, and
then posted to those respective target language discussion groups. The
collection and
posting of the messages will be accomplished by the Message Collection/Posting
Software which forms a part of the present invention. The new messages which
are
collected on a periodic basis are sent to a commercially available Machine
Translation
(MT) software for translation.
Before the input text is actually submitted to the MT software, the input text
is
passed through a filter software program which preprocesses the data before it
is
submitted to the MT software. The filter identifies and marks strings which
are best left
untranslated by the MT software; such as personal names, company product
names, file
and path names, commands, samples of source code, and the like. By marking
these
strings, the filter notifies the MT software to leave those strings
untranslated. These
strings are then linked to a preceding "hookword". Hookwords are automatically
inserted then deleted in post-processing and are contained in dictionaries
with a part-of
speech and other grammatical features to effect rearrangement of the word in
the target
language. Once the translation process is complete, the translations are
collected and
posted, by the Message Collectiorv'Posting Software, to the target language
discussion
groups at the same location within the message structure as the original
version of the
message.
4
CA 02544238 1996-02-14
Thus, some of the messages on a French version of a discussion group will have
been originally written in French whereas many will be translations of
messages which
were originally written in a different language. Each of the messages
translated from
another language version of the parallel discussion group may contain both the
translated
message and the original language text. For example, a message originally
written in the
English version of the discussion group and translated to the French version
may be
shown on screen with the original English text in addition to the French
translation.
Therefore, the method of the invention creates structure and process to enable
discussion group users, of different languages, to communicate with one
another.
Further features and advantages of this invention will become apparent when
viewed in
light of the following drawings and detailed description.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 illustrates a Bulletin Board System that is accessed by users that
speak
different languages;
Figure 2 is a diagrammatic view of one embodiment of the present invention, in
which multiple languages are in communication over parallel discussion groups;
Figure 3 illustrates the overall message translation process of the present
invention;
4a
CA 02544238 1996-02-14
Figure 4 is a diagrammatic view of an example of one embodiment of the present
invention;
Figure 5 is an example of a user's computer screen display while accessing the
present
invention through CompuServe;
s Figure 6 is an example of a screen display of a user entering a particular
discussion
group;
Figure 7 is an example of a screen display of a user accessing a parallel
discussion
group via point and click;
Figure 8 is an example of a screen display of an original language message and
a
1 o parallel foreign language translation message;
Figure 9 is an example of a screen display for a user creating an original
message to be
posted on a parallel foreign language discussion group;
Figure 10 is an example of a screen display for a user responding to the
original
message shown in Figure 9;
i s Figure 11 is a diagrammatic view of the process of the present invention;
Figure 12 is a flow chart of the machine translation software process flow of
the
present invention; and
Figure 13 A-D is a detailed flow diagram of the process shows in Figure 12.
DETAILED DESCR»TION OF PREFERRED EMBODIMENTIS)
2 o Referring now to Figure l, a bulletin board system is shown. Users 20, 22,
24, 26 can
access the bulletin board system 28 through a personal computer 30 and a
modem. The users'
personal computers are ordinarily connected to the bulletin board by telephone
Iines 32. The
bulletin board system is subdivided by subject discussion areas; for example
sports, travel,
skiing, stamp collecting 34, 36, 38, 40 respectively. These discussion areas
may be further
CA 02544238 1996-02-14
broken down into several sections based on subject. For example, a sports
discussion area
may be separated into football, basketball, and baseball sections. Once a user
obtains access to
a particular bulletin board system, he or she may then log on to any of the
discussion areas
contained in the particular bulletin board system
s Figure 2 illustrates the basic structure of parallel discussion groups,
according to the
present invention. Figure 2 illustrates a communication system which is
comprised of three
parallel discussion groups in English, French, and German, respectively shown
at 42, 44, 46.
The number of parallel discussion groups for a particular system varies with
the number of
languages it is designed to accommodate. For example, a parallel system
designed to
1 o accommodate ten di$'erent languages will have ten parallel discussion
groups; one for each
language.
Figure 2 illustrates the message structuring of one embodiment of the present
invention. Upon logging onto a discussion group, the user may choose his or
her language
preference. For example, a French user may log onto a discussion group and
access the
l s French version of the system For example, Figure 2 depicts the message
structuring of a
French version of a parallel discussion group. The message board section names
and messages
will be in French. Each of the messages translated from other language
versions may also
contain the text in the original language in addition to the French
translation. If the French
user posts a message to the parallel discussion group as shown in Figure 2, it
will be translated
2 o to English and German and posted to the English and German versions of the
system
The content, structure, and "thread" structure of the parallel discussion
groups are
preferably identical. When a message is posted and another user responds to
it, a "thread" is
created. A thread is a string of two or more related messages that are grouped
together and
preserve the original subject heading. For example, a French reply to an
original English
6
CA 02544238 1996-02-14
message is assigned properly to the English message. In a typical BB S, a user
can display a
hierarchical view of a thread to examine the history of the thread and the
relationships between
messages posted to that discussion group.
Figure 3 depicts the overall process steps involved in translating and posting
a message
from its original version to the other language discussion groups. Every
message posted to a
discussion group is preferably assigned a message identification number by the
discussion
group software. In the preferred embodiment of this invention, the Message
Collection/Posting Software 48 identifies new messages posted to a discussion
group by
comparing the message identification numbers to a continuously created
database of old
i o message identification numbers. New messages are collected on a periodic
basis and
submitted to the filter software 50 for preprocessing of the message text
before submitting it
to the MT Software 52.
The filter identifies and marks strings in the message text which are best
left
untranslated; such as, personal names, company and product names, quoted text,
file and path
i 5 names, commands, and samples of source code. These strings are then linked
to a preceding
hookword. Hookwords are automatically inserted then deleted in post-processing
54.
Hookwords are contained in the Intergraph dictionaries with part-of speech and
grammatical
features which affect rearrangement of the nontranslated word in the target
language. Once
translated, the message is delivered to the target discussion group 56 in the
language of that
2 o discussion group.
In the preferred embodiment, direct translation of the message text is
performed (e.g.
direct French to English translation). The MT Software analyzes and tags the
sentences of the
message text using Linguistic algorithms. The words in the sentences are then
looked up in
electronic dictionaries to identify their translations. The translated words
are then reordered
7
CA 02544238 1996-02-14
and inflected according to the rules of the target language. Machine
translation software has
been developed by the Intergraph Corporation and is commercially available.
Customized .
dictionaries containing computer specific terminology were developed by
CompuServe
Incorporated.
s Once the translation process is complete, the translations are collected and
posted to
the target language discussion group at the same location within the message
structure. This
is accomplished by the Message Collection/Posting Software.
The following example will serve to provide a more detailed description of the
present
invention. It must be stressed that the following is only an example
implementation of the
i o invention and is not intended to limit the scope of the invention to the
specific features
described in the example.
EXAMPLE
The MacCIM Forum~ discussion group is one of the most popular and widely-used
i s groups on the CompuServe Network, especially in Europe. New and
experienced
CompuServe Information Service ("CIS") users access MacCIM to learn how to use
the
service, to enhance their access to the CIS Network, and to o$'er suggestions
for improving
MacCIM. French and German users frequently log on to this discussion group,
but until
recently only minimal support was available to them in their native languages.
Refer to Figure
2 0 4 for a high-level overview of this embodiment of the invention, known as
the MT Model.
The MT Model has three views that are each separate CompuServe discussion
groups
-~ one in English (EMCIMSU) 58, one in French (FMCIMSU) 60, and one in German
(DMCIMSU) 62. A user may only see one view at a time. All three discussion
groups
contain the same message content and configuration. In this Example, the
following
8
CA 02544238 1996-02-14
translation directions occur: English to French; English to German; French to
English; and
German to English.
In all three discussion groups, the message board section names appear in the
native language. Messages in each section appear in the native language or are
translated from
s another language. All of the discussion groups in this embodiment have an MT
Comments
section that is designated for any issues relating to machine translation.
When a user posts a
message on any one of the three discussion groups, the message translation
process performs
the translations and posts the translated messages onto the other two
discussion groups.
Messages translated from another language preferably contain the translation
as well as the
1 o message as it originally appeared.
There are at least two ways a user can access the discussion groups. The first
is
to type GO MACCIM from the CompuServe menu. The discussion group appears in
the
language that is defined in the user's language preference setting. A user
whose language
preference is set to Engtisb views EMCIMSU. A user whose language preference
is set to
1 s French views FMCIMSU. A user whose language preference is set to German
views
DMCIMSU.
The second way to access any of these discussion groups is by typing GO plus
any
one of the three translation discussion group names. For example, if a French
user types GO
EMCIMSU, the English translation discussion group would display even though
the user's
20 language preference is set to French. This French user would still be able
to view and post
messages in this discussion group.
An English-speaking user logs on to MacCIM to view new messages. Because she
typed GO MACCIM and because her language preference in her CompuServe profile
is set to
English, the system navigates her to EMCIMSU. She sees a screen similar to
that shown in
9
CA 02544238 1996-02-14
Figure 5. When the user selects the MacCIM Forum~ discussion group, the screen
displays
as shown in Figure 6. Next she selects the Messages icon to display the
message board. As
shown in Figure 7, the user selects new messages in the MT Comments section,
which refers
to issues relating to machine translation. She notices she has two messages in
this section.
s The user reads a message translated from French to English as shown in
Figure 8. In
this embodiment, the message includes both the translation and the text as it
was originally
written in French. During a MacCIM session, the English-speaking user decides
to post a
message to a person who she knows has recently joined the MacCIM Forum~. She
posts the
message to the General Questions section as shown in Figure 9. Later she logs
off. Because
i o this user is also fluent in French, she decides to bypass the English
version of the MacCIM~
Forum~ and go directly to the French version to view any responses to her
message. She
finds she has received one response as shown in Figure 10.
The user may still view any responses to her message by staying on the English
version of the discussion group because any French or German responses to her
message
i s would still be translated and posted there. The present invention does not
require her to go to
the French group to view French replies.
Continuing with this Example, certain terms are defined to facilitate an
understanding
of the invention. Each translated message has two banners: one that introduces
the translated
message and another that introduces the message as it appeared in its original
language.
2 c DP/Translator is the Intergraph~ Corporation's natural language
translation system used to
perform the MT Model's language translation in this embodiment of the
invention. Filtering
refers to the process by which messages are prepared for translation by
DPlTranslator. A
header is the CompuServe message information that appears at the top of any
standard
message.
CA 02544238 1996-02-14
In this embodiment of the invention, message translation is a batch process
that runs
on an hourly basis. It is driven by a Unix Shell script tricycle, which calls
all the necessary
programs in the order they need to run. A language code is a one-letter code
that refers to a
language. For example, French is referred to as f. In the following detailed
process
s descriptions, this is abbreviated as Lang code. A language pair is a two-
letter code that refers
to the two languages from and to which translation occurs. The first letter is
the source
language; the second is the target language. For example, in an English-to-
German
translation, the language pair is eg. In the detailed process descriptions,
this is abbreviated as
long pair. Post-filtering refers to the process by which messages are further
prepared for the
i o pre-posting process once they have been translated. A shell script is a
Unix-based file that
contains a list of Unix commands.
A Unix shell script tricycle runs at a predetermined interval (in one
preferred
embodiment this interval was set at one hour) to retrieve new messages that
users post onto
the discussion groups. It passes them through the translation process and
posts them to all
1 s three parallel discussion groups, in this Example.
Referring to Figures 11 and 12 a user posted message can originate from any of
the
three parallel discussion groups. The MT Message Read Process reads the
message from the
discussion group. A body text file is created as input to a filtering process.
The message
passes through the translation process. English is translated to French and
German. German
2 c or French is translated to English.
The message is filtered again to remove the markers added during the MT Filter
process for text that was not intended for translation. During translation,
the original message
text was saved in the originating language to be included with the
translation. The translated
message is further prepared for posting to the other two discussion groups. A
posting output
11
CA 02544238 1996-02-14
file is created that contains translated text, text in the originating
language, and additional
information required for posting.
MT Message Number Database Update adds a line of three message numbers (in
this
Example) to a database to maintain a cross reference of message identifiers
among the three
s discussion groups. These include the message number that the source
discussion group
assigns to the message and the corresponding message numbers for the other two
destination
discussion group s.
The following introduces the software components of the MT process. The
diagram
in Figure 12 shows the flow in which these components nm.
Name Type Description
tricycle Unix shell script Orchestrates the hourly cycle of the MT process by
calling
each program in their appropriate order
rJ'or C program Reads messages from MacCIM discussion groups,
determines whether they need to be translated, and creates
a message header file and a message body text file; called
by tricycle
rtrans Unix shell script Calls fm to filter the message file, DPlTranslator to
translate the messages, and ufm to perform post-translation
filtering; called by tricycle
fm C program Identifies text that is not to be translated and surrounds
with markers; called by rtranr
DPlTranslator Intergraph product Performs message translation; called by
rtrans
uJm C program Removes markers that surround text identified by fm as
non-translatable; called by rtraru
Apart C program Creates a new file containing translated message to post to
discussion groups; called by tricycle
w, for C program Posts messages to their corresponding discussion groups;
called by tricycle
12
CA 02544238 1996-02-14
umdb C program Updates a message number database; called by tricycle
The shell script tricycle governs sequencing of the MT process shown in
greater
detail in Figures 13A-13D. Preferably, each hour, it runs the MT programs in
the appropriate
order and for the corz-ect number of times.
s The tricycle script does the following.
( 1 ) Sets some shell variables based on the current date and time; these are
used to
construct unique names for the temporary files that will be created;
(2) Enters a request in the at queue to cause tricycle to be run again at half
past
the next hour;
to (3) Backs up a message number database;
(4) Calls the program rfor for the English discussion group; rfor connects to
CompuServe and reads new message headers to identify new messages to be
translated;
(5) Calls the shell script rtrans for the English discussion group; rtrans
runs the
messages through a filter program fin that identifies untranslatable message
text, sends the
i s message to DPlTranrlator for translation, and routes the output from the
translation to a post
filtering program ufrn;
13
CA 02544238 1996-02-14
(6) Calls the program ppost for the English discussion group; ppost creates
the file
to be posted to the discussion groups; this program is run twice, once to
create the English-to-
German translation file and another to create the English-to-French
translation file;
(7) Performs steps 3 through 5 for the French and German discussion groups,
s although it only runs rtrans once because translation only occurs for French
to English and
German to English;
(8) Calls the program wfor six times, once for each language pair (ef, eg, fe,
Jg,
ge, ~, to post all translated messages to their appropriate discussion groups;
and,
(9) Calls program umdb six times, once for each language pair (ef, eg, fe, fg,
ge,
i o ~, to update message number database.
The program rfor reads messages from the discussion group, checks to make sure
they are not machine translated, and prepares them for filtering. The shell
script tricycle calls
rfor and provides it with the necessary parameters.
The program rfor does the following:
i s ( 1 ) Connects to CompuServe;
(2) Reads new message headers from the discussion group;
14
CA 02544238 1996-02-14
(3) Compares the message numbers in the headers against those listed in a
message number database to determine whether the message is a machine-
translated message
from another discussion group; if so, it skips the message and goes to the
next one;
(4) Creates a message header file that contains the message number, a date and
time stamp, subject, sender, and recipient information; and
(5) Creates a message body text file that contains the message number,
subject,
and message text.
The following parameters are specified:
discussion name of discussion group from which message originated
group-name (emcimsu, fmcimsu, dmcimsu)
lung-code one-letter code (e, f, or g) specifying the source language
specified in the CompuServe user profile
msg-dbase name of the Unix database that holds a cross reference of all
message numbers for each discussion group
hdr-file name of the header file created as output
msg-file name of the message file created as output
Inputs to rfor are discussion group-name, lang-code, and msg-dbase provided by
i o tricycle. Outputs from rfor include hdr-file that is input to the MT Pre-
Posting Process, and
msg-file that is the input to the MT Filter Process.
The shell script rtrans calls the programs that filter, translate, and
refilter the
messages to be translated. The shell script tricycle calls rtrans.
CA 02544238 1996-02-14
The program rtrans does the following:
( 1 ) Takes as input from tricycle the base name (referred to as base-name in
some
of the program descriptions that follow), the source language code, and the
target language
code;
s (2) If the source language code is g for German, then calls the filter
program fm
with a special -g parameter that handles capitalization grammar rules specific
to the German
language;
(3) If the source language code is f or e, calls filter program fm without
special
capitalization rules;
i o (4) Calls DPlTranslator, providing it with the output file from fm and
additional
parameters; and,
(5) Retrieves the output and passes it through the post filter program ufm.
The program fm runs the message file created during the MT Message Read
Process through a filtering process to prepare the message text for
translation. The shell script
15 rtranr runs the program and provides it with the necessary parameters. Then
it submits the
filtered output file as input to the translation program DPlTranrlator.
The program fm does the following:
( 1 ) Evaluates text for strings that should not be translated;
(2) Places marks around the identified strings; and,
16
CA 02544238 1996-02-14
(3) Ifthe source language is German, performs special filtering to handle
nouns
that are always capitalized.
When it runs fm, rtrarar redirects input from the file base-name. source-fang.
When it runs fm, rtranr directs output to a target file name called base-
name.i.
The program ufm runs the file generated from the language translation process
DPlTranrlator through a process that removes markers surrounding text strings
that were not
translated. The shell script rtrarrs runs the program and provides it with the
necessary
parameters.
1 o When it runs ufm, rtrans redirects input from the file base name.o, which
is an output
file generated by DPlTrarrslator during the translation process. When it runs
ufm, rtrans
directs output to a target file base-name.target-Lang, where target-fang is
the one-character
language code (e, f, g) of the language to which the message was translated.
The program ppast creates a new file to post to the discussion groups. The
shell
script tricycle calls the program six times, once for each source-target
language translation
pair (ef, eg, fe, fg, ge, ~. The header and message files may contain more
than one message
per run of ppart.
The program ppast does the following:
( 1 ) Retrieves the header file created during the MT Message Read process;
(2) Retrieves the original message text written in the source language;
(3) Retrieves the translated message text;
(4) Determines the translation direction; and,
17
CA 02544238 1996-02-14
( 5 ) Creates an output file that contains the entire uanslation message. The
following parameters are specified:
hdr-file header file created during the MT Message Read Process
orig-file message in its original form, created during the MT Message
Read Process
trans-file translated message file
lung-pair two-character code identifying the source-target language
translation pair
post-file file containing entire message to be posted onto the target
discussion group
Inputs to ppost include the header file from the MT Message Read Process (base-
s name.hdr), the original text file from the MT Message Read Process and, the
source-target
language pair. The posting files that ppast creates contain: the original
message number; the
parent message number if this is a reply; sender, recipient and (translated)
subject information;
a banner that identifies the message to follow as translated text and displays
the original
subject; the translated text; a banner that identifies the message to follow
as the original text;
z o and, the original text.
The program wfor accesses CompuServe and posts translated messages to their
corresponding CompuServe Forum~ discussion groups. The shell script tricycle
calls the
program six times, once for each source-target language translation pair (eJ,
eg, fe, fg, ge, g~.
The program wfor does the following:
1 s ( 1 ) Accesses the appropriate CompuServe discussion group;
18
CA 02544238 1996-02-14
(2) Pasts the message files created during the MT Pre-Posting Process; and,
(3) Appends message and message number information to a log file that is later
used to update the message number database.
The following parameters are specified:
s discussion group-name name of destination discussion group
post-file name of the file created during the MT Pre-Posting
Process
tang-pair two-character code identifying the sowce-target
language translation pair
1 o msg-dbase name of the Unix database that holds a cross reference
of all message numbers for each discussion group
log-file file to which message numbers are appended and from
which the cross reference of message numbers are
updated
1 s Inputs to wfor include discussion group-name, post-file, fang-pair, and
msg-dbase
provided by tricycle. The outputs of wfor include the updated discussion
groups and six
message database log files, one for each language pair.
The program umdb updates the message number database with the message IDs
assigned to each message on a discussion group. The shell script tricycle
calls umdb six times,
20 once for each source-target language translation pair (ef, eg, fe, fg, ge,
gf).
19
CA 02544238 1996-02-14
Umdb does the following:
( 1 ) Opens the message database; and,
{2) Loads the message data into the database.
The following parameters are specified:
log-file file to which message numbers are appended and from which the
message number database is updated.
lung-pair two-character code identifying the source-target language
translation pair.
msg-dbase name of the Unix database that holds a cross reference of all
message numbers for each discussion group.
s Inputs to umdb include log-file, fang-pair, and msg-dbase provided by
tricycle. The
output of umdb is the updated database.
END OF EXAMPLE
