Managing enterprise terminology

1
Managing enterprise terminology

• Tools demo
• Terminology Workshop
• Kara Warburton
• San Francisco, 2008

2
Terminology tools

• SDL Trados Termbase and MultiTerm
• Star TermStar
• Multicorpora MultiTrans
• Alchemy Catalyst
• iTerm
• Interverbum TermWeb
• IBM TransLexis (proprietary)
• Term extraction functions

3
SDL Termbase

• Part of SDLX
• All fields are at the term level
• Fixed number of synonyms and related terms
  allowed

4
SDL MultiTerm

• Full-fledged terminology management system
• Feeds auto-lookup function (active term
  recognition) in Translators Workbench
• MSWord editor
• TagEditor
• Terminology consistency checker
• Client/server environment or standalone
• Available Web interface MultiTerm online
• Role-based user classes

5
MultiTerm

• Highly-customizable entry model (in some respects
  this is a disadvantage)
• Direct access from editors insert term from db
  into text, add term to db
• MSWord plugin
• Click to add unfound term as candidate
• Input models preset mandatory fields and
  default values
• Advanced search functions and filters
• Customizable input models facilitate and
  control data entry

6
MultiTerm

• Termbase management functions
• Find entries with the same term
• Find incomplete entries
• Import
• Decisions need to be made on a file basis, not
  possible on an entry basis (add, skip, overwrite,
  merge)
• Can be carried out from MultiTerm Online
• File conversion utilities

7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
MultiTerm field types
12
MultiTerm issues and limitations

• Auto term recognition can slow down performance
  of the TM lookup in the Workbench (Translators
  reaction turn it off!)
• Auto term recognition can only use one termbase
  at a time
• License based on number of users
• Poor management of doublets
• Difficult to make global changes (export, change,
  reimport)

13
MultiTerm issues and limitations

• Import functions are file-based not entry-based
• Poor management of cross-references
• No relational or hierarchical fields
• User interface for auto-lookup when there are
  multiple translations
• Only the default database is available for
  editing

14
STAR Transit Termstar

• STAR Group is the biggest privately-held
  localisation service provider world-wide.
• All STAR Group offices use Transit XV as the TM
  tool
• TermStar is the terminology function that works
  with Transit
• Transit is the only translation memory system
  that has a file-based structure. Translation
  memories are stored as text-based XML files.
• Column-based translation interface

15
TermStar

• WebTerm Web interface to TermStar
• Connect via LAN to termbase from Workbench
• Plugin for MSWord and Framemaker search for
  terms and add terms to the db directly from the
  editor.
• Field-based global changes.
• Search memories for terms in term db.

16
TermStar - import

• More advanced import functions than any other CAT
  tool
• Interactive import at entry level skip, append,
  add, merge
• User-specific definition of duplicates based on
  selection of fields for comparison
• Good logging functions
• Ability to preset default values

17
(No Transcript)
18
(No Transcript)
19
TermStar issues and limitations

• Cross-referencing
• Lack of term autonomy for variants
• Lack of specific translation pointer
• Limited entry customization (in some respects
  this is an advantage)
• Import of multi-occurrence elements
• No field-level export customization

20
TermStar issues and limitations

• Synonyms

• Lack of relational fields, and hierarchical
  relations

21
MultiCorpora MultiTrans

• TextBase TM CAT tool
• Creates aligned translation memories from
  parallel texts (TextBase)
• Insert terms/translations from textbase to
  termbase
• Translation Agent
• Plugin for MSWord, PowerPoint, and Word Perfec
• Terminology checker
• Possibility to have separate lookup and output
  termbases
• Various automatic replacement options for terms
  from termbase and for segments from textbase
• Analysis agent pre-identify important terms to
  translate

22
MultiCorpora Termbase

• Terminology function of MultiTrans
• Customizable term entry form
• Can search multiple databases at once
• Search by field
• Search filters
• Web interface available

23
MultiCorpora Termbase

• Easy to make global changes in entire db or
  selected entries
• Cross-references are non-ambiguous
• Hierarchical fields
• Bidirectional hyperlinks
• Identification and management of duplicates
• Identification and management of broken links

24
(No Transcript)
25
(No Transcript)
26
MultiCorpora Termbase field types
27
MultiCorpora Termbase issues and limitations

• Fields cant be repeated on the fly (have to
  pre-specify number)
• Relation fields are not bidirectional
• TBX not yet supported
• Minor usability problems
• No interactive import identification and
  management of duplicates after import

28
Alchemy Catalyst

• Very sophisticated CAT tool for localization of
  software files
• Supports many file types
• Complementary to Trados SDL so can connect to
  MultiTerm
• No terminology management function
• Allows TVT to be carried out in real time
• Glossary consistency check

29
Catalyst issues and limitations

• Does not support sub-segment analysis
• glossary is just another kind of TM

30
Interverbum TermWeb

• Used by LISA Terminology SIG
• Entirely Web-based
• Standalone system. Not linked to any CAT tool.

31
iTerm

• Copenhagen School of Business
• Stand-alone terminology management system not
  connected to any CAT tool
• Used by ISO TC37

32
iTerm relational fields
33
Term Extraction

• MultiTerm Extract
• SDL Phrasefinder
• Multicorpora
• Temis (good tool, no demo)
• IBM TermExt (proprietary)

34
MultiTerm Extract

• Statistical extraction (based on frequency)
• Works on all languages
• Extract terms from monolingual texts or bilingual
  aligned texts (memories)
• Export results to MultiTerm
• Many file formats supported

35
PhraseFinder

• Linguistic extraction (grammar rules and lexical
  databases)
• Limited languages English, French, German,
  Italian, Dutch, Spanish, Portuguese
• Fewer file types supported for bilingual
  extraction, only SDLX memories are supported.

36
Phrasefinder weaknesses

• No lemmatization
• Multiple term candidates of same term from same
  context
• Prepositions appended to terms
• Term identification errors
• Crashes on large documents

37
MultiCorpora terminology extractor

• Statistical based
• Monolingual or bilingual
• Stopword lists
• Easy to edit the output and send to term db

38
Terminology databases

• IBM LOGOS
• Onterm (Ontario Government)
• Termium (Canadian Government)
• IATE (formerly Eurodicautom)
• BTQ (Quebec Government)
• Danterm