Download Carabao DIY Machine Translation Kit

Size: 2.79 MB
License: Freeware
OS: Win95,Win98,WinME,WinXP,WinNT 3.x,WinNT 4.x,Windows2000,Windows2003
Developer:Vadim Berman (» more programs)
View: Antivirus Report

Download Now

Add to Download Basket

Program system requirements: Windows 9x or Windows NT/2000/XP

New in this version: Version 0.42
ADDED
* Import of idioms
* 'Suppress value' field in measures / metrics which will allow to omit specified numeric values (for example, to convert numbers to text, etc.). It is currently not active yet.
IMPROVED
* Menu in data manager
FIXED
* Importer bugs
* Resize handling in some screens

Version 0.41
ADDED
* Capability to set a limit for imported lines
IMPROVED
* Importer performance & GUI
FIXED
* More importer bugs

Version 0.40
ADDED
* Looks like the basic French and English morphology is ready
* System messages which are inserted into content on certain occations. Currently they are active before and after the content
* Style units granted by modifiers
* Fixed Increment in measure conversion; enables all linear conversions without resorting to formulas
IMPROVED
* Lexical Unit Sheet does not attempt to re-generate the canonic form when confirming the form anymore
FIXED
* Importer parsing bugs
* Morphological tagger flaws with phonetic compatibility
NOTE:
Due to the structural changes in data tables, your old data might not work with the new version. Email us for instructions how to update your database.

Version 0.39
ADDED
* Finally! A morphological tagger based on uninflected form recognition. It is currently used in mass import - no more "conditional text", just link your tag to designated rule units and the rest is done by modifiers. The tagger is also linked to data entry. Currently the configured rules allow recognizing French and English nouns.
FIXED
* Jumps to incorrect record when using locator in multilingual tables

Version 0.38
ADDED
* Capability to hold modifiers that are preserved in canonic form. This will be later used in wordlist tagging & recognition
IMPROVED
* Formatting for syntax delimiters by ability to specifying where spaces should be inserted or omitted
FIXED
* Situation with sequences when the same sequence is found twice or more times in the same sentence
* Phonetic compatibility check

Version 0.37
ADDED
* Support for compound words and languages that don't use white spaces, such as Japanese, Thai, etc.
* Inside Carabao document descibing the architecture of Carabao
IMPROVED
* Canonic form generator now verifies the rule units
FIXED
* Uninflected form generator
* Phonetic compatibility check

Version 0.36
ADDED
* "Late sequences" - sequences detected on the target language entries. So far Carabao was crippled by inability to select sequences based on target rather than source language lexical units. For example, to properly translate to English comparative or superlative adjective forms, one needs to know how a respective English adjective is formed (by attaching -er / -est or using "more" / "most"). With "late sequences", it is possible to create a "gateway" sequence which will "export" a member's rule units and then detect where to go from there
* Capability to enter words in a regular uninflected form, rather than Carabao canonic form
* Option to pull rule and style units in sequence member editor
FIXED
* Nasty rearrangement bug in sequences having "not" and "gap" members in front

Version 0.35
IMPROVED
* Tokens table now works in two separate modes, lexical units and sequences
ADDED
* "Active" flag for languages which controls whether the languages will be displayed in the translation console

NOTE:
Due to the structural changes in data tables, your old data might not work with the new version. Email us for instructions how to update your database.

Version 0.34
FIXED:
* Rearrangement bug when adding an element with a new identity
IMPROVED
* Sequence descriptions now filter descriptions of tokens by rule units
ADDED
* French language rules for the main database. Looks like we'll have a real world database here, folks!

Version 0.33
FIXED:
* Inability to identify tokens in a language without modifiers (maximum modifer length = 0)
ADDED:
* Capability to display lexicon usage statistics of the translated content

Version 0.32
ADDED:
* Capability to display style and subject statistics (Debug tab in Translation console)
IMPROVED:
* Semantic link table and details are reset to currently active language rather than the first matching one
* Changes in manual, including a chapter on mass import

Version 0.31
FIXED:
* Thesaurus update mode
* Measure conversion examples
ADDED:
* More French grammar rules
* Fields for designer's notes in the tables of modifiers and languages
* Ability to complement existing database by incremental import.
* Ability to convert measure by formula, for example temperature - Celsius and Fahrenheit
IMPROVED:
* Terms fixed to standard ones
* User's guide
* Field for notes in the table of tokens enlarged
* Selection of proper record from a family of tokens
* In Tokens table, the column holding a token's text now expands or contracts on resize for better viewing.
NOTE:
Due to the structural changes in data tables, your old data might not work with the new version. Email the author for instructions how to update your database.

Version 0.30
FIXED:
* Words mass import actually starts to work
ADDED:
* Some measure conversion examples for the default database
* Validation for missing rule units in import mode - no need to guess what's missing
* Unidirectional import mode (without a need for a second validation dictionary)
* Capability to display a message when starting the data manager
IMPROVED:
* User's guide. It is still full of inconsistencies and errors, but is getting fixed gradually

Version 0.29
Carabao DIY Machine Translation Kit released

[Zoom screenshot]

[Editor's Award]

Publisher's Description:

Carabao Do-It-Yourself Machine Translation Kit is a fully customizable machine translation and conlangery toolkit. It allows creation of your own machine translation interlingua-modeled databases and using them on-the-fly. Unique capabilities of Carabao and effective data entry allow using it as a testbed for linguistic models and creation of specialized dictionaries. Features include:
* Faster and more nuance-tolerant word entry.
* Style sensitivity and control - ability to reshape the result so it will bear the same meaning but in different style. For example, the user may write a phrase using common words in French, and the program would translate it into official style in English.
* Model and entities are grammar-independent and are suitable for any language without the need to reprogram or even recompile. Lexical data is stored in transactional database.
* Interlingual approach to storing lexical data.
* Descriptive mini-language for collocations, rules and idioms that allows effective handling of lexical holes.
* Inline conversion of measures.
* Built-in customized instant dictionary printing capability.
* Built-in thesaurus for words and idioms found in the processed text.
* Ability to trace the translation process.

Comments

No Comments have been Posted.

Post Comment

Alerts

Carabao DIY Machine Translation Kit Program Releases History

Software:	Carabao DIY Machine Translation Kit 0.41
Update date:	0000-00-00 00:00:00
Release Notes:	Version 0.41 ADDED * Capability to set a limit for imported lines IMPROVED * Importer performance & GUI FIXED * More importer bugs Version 0.40 ADDED * Looks like the basic French and English morphology is ready * System messages which are inserted into content on certain occations. Currently they are active before and after the content * Style units granted by modifiers * Fixed Increment in measure conversion; enables all linear conversions without resorting to formulas IMPROVED: * Lexical Unit Sheet does not attempt to re-generate the canonic form when confirming the form anymore FIXED: * Importer parsing bugs * Morphological tagger flaws with phonetic compatibility NOTE: Due to the structural changes in data tables, your old data might not work with the new version. Email us for instructions how to update your database. ADDED * Finally! A tagger based on uninflected form recognition. It is currently used in mass import - no more "conditional text", just link your tag to designated rule units and the rest is done by modifiers. The tagger is also linked to data entry. Currently the configured rules allow recognizing French and English nouns FIXED * Jumps to incorrect record when using locator in multilingual tables Version 0.38 ADDED * Capability to hold modifiers that are preserved in canonic form. This will be later used in untagged wordlists tagging & recognition IMPROVED * Better formatting for syntax delimiters by ability to specifying where spaces should be inserted or omitted FIXED * Situation with sequences when the same sequence is found twice or more times in the same sentence * Phonetic compatibility check Version 0.37 ADDED * Support for compound words and languages without that don't use white spaces, such as Japanese, Thai, etc. * Inside Carabao document descibing the architecture of Carabao IMPROVED * Canonic form generator now verifies the rule units FIXED * Uninflected form generator * Phonetic compatibility check Version 0.36 ADDED * "Late sequences" - sequences detected on the target language entries. So far Carabao was crippled by inability to select sequences based on target rather than source language lexical units. For example, to properly translate to English comparative or superlative adjective forms, one needs to know how a respective English adjective is formed (by attaching -er / -est or using "more" / "most"). With "late sequences", it is possible to create a "gateway" sequence which will "export" a member's rule units and then detect where to go from there * Capability to enter words in a regular uninflected form, rather than Carabao canonic form * Option to pull rule and style units in sequence member editor FIXED * Nasty rearrangement bug in sequences having "not" and "gap" members in front Version 0.35 IMPROVED * Tokens table now works in two separate modes ADDED * "Active" flag for languages which controls whether the languages will be displayed in the translation console NOTE: Due to the structural changes in data tables, your old data might not work with the new version. Email the author for instructions how to update your database. Version 0.34 FIXED: * Rearrangement bug when adding an element with a new identity IMPROVED * Sequence descriptions now filter descriptions of tokens by rule units ADDED * French language rules for the main database. Looks like we'll have a real world database soon, folks! Version 0.33 FIXED: * Inability to identify tokens in a language without modifiers (maximum modifer length = 0) ADDED: * Capability to display lexicon usage statistics of the translated content Version 0.32 ADDED: * Capability to display style and subject statistics (Debug tab in Translation console) IMPROVED: * Semantic link table and details are reset to currently active language rather than the first matching one * Changes in manual, including a chapter on mass import Version 0.31 FIXED: * Thesaurus update mode * Measure conversion examples ADDED: * More French grammar rules * Fields for designer's notes in the tables of modifiers and languages * Ability to complement existing database by incremental import. * Ability to convert measure by formula, for example temperature - Celsius and Fahrenheit IMPROVED: * Terms fixed to standard ones * User's guide * Field for notes in the table of tokens enlarged * Selection of proper record from a family of tokens * In Tokens table, the column holding a token's text now expands or contracts on resize for better viewing. Version 0.30 FIXED: * Words mass import actually starts to work ADDED: * Some measure conversion examples for the default database * Validation for missing rule units in import mode - no need to guess what's missing * Unidirectional import mode (without a need for a reverse validation dictionary) * Capability to display a message when starting the data manager IMPROVED: * User's guide. It is still full of inconsistencies and errors, but is getting fixed gradually