Exact Equivalence (EE)

Used where two terms have identical meanings in both the source and target languages (identical concept sets).

Inexact Equivalence (IE)

Used where the source and target language terms do not have exactly the same meaning but express the same general concept (overlapping concept sets).

Broader Equivalence (BE)

Used where the translation is close but implies a slightly broader meaning (superset/subset relationship between concept sets).

Narrower Equivalence (NE)

Used where the translation is close but implies a slightly narrower meaning (subset/superset relationship between concept sets).

A revised user guide is now available which describes the multilingual features.

• Clicking the “Download” button above downloads the latest version of Cognatrix. This includes the updated user guide which can be accessed from the Help menu.
• Clicking the “Manual” button above downloads just the revised user guide (PDF).

defining languages

Cognatrix ships with 20 language codes defined (those for the European Union). New languages can be added, deleted or edited using the Languages... command in the File menu. Each language definition includes a two- or three-character code, a language name and a native language name. The code and language name are typically expressed in English whereas the native language name is typically expressed in the script of the language. There is, however, no need to conform to this approach as all fields are completely customisable. There is no limit on the number of languages that may be defined.

(see also http://en.wikipedia.org/wiki/List_of_ISO_639-2_codes)

language codes

Each item can now have a language associated with it. Language codes are selected via popup menu in either Tree or List view. List view can also be sorted by language.

language-equivalence relationships

Four new relationship types are defined: Exact Equivalence, Inexact Equivalence, Broader Equivalence and Narrower Equivalence. Any time two terms (preferred or variant) have different language codes, they are candidates for forming relationships of these types. Relationship types of Multiple Equivalence and No Equivalence are implicit. The former occur when a term participates in multiple language-equivalent relationships and the latter when a term does not participate in any language-equivalent relationships. There is no limit on the number of language-equivalent relationships that a term may be involved in.

inspector window

Language-equivalent relationships appear in the inspector window and in printouts of thesaurus items.

XML export

Cognatrix XML now includes the “xml:lang=” attribute to describe each item’s language code. Zthes XML makes use of the <language> tag for the same purpose. XML generated in MultiTes format does not currently include language support. All thesaurus data containing Unicode is properly encoded to parse correctly in XML.

HTML export

Three new substitution keys have been defined: QIDLanguageCode, QIDLanguageName and QIDLanguageNativeName to support the display of language information in generated pages. All thesaurus data containing Unicode is properly encoded to render correctly in HTML.

Cognatrix DTD 1.3

The new relationship types imply changes to the DTD which has now been updated to version 1.3.

HTML generation

HTML generation currently assumes that terms can be divided into the 26 letters of the English alphabet plus numbers and “other”. This may not be appropriate where non-Roman scripts are involved.

printing the thesaurus tree

The index that is generated when the taxonomy is printed makes the same assumption as HTML generation. In addition, the one-line summary of each item does not describe its language. Also, variant terms that are only involved in language-equivalence relationships are described, incorrectly, as orphan terms.

suspect multilingual relationships

Like most relationships created using Cognatrix, language-equivalent relationships are implemented as guidelines. For example, it is only possible to create a language-equivalent relationship where the two terms have different language codes. However, if the language code of a term is changed after a language-equivalent has been formed, this “rule” may be broken. In such a situation, Cognatrix flags the issue with an alert symbol. At present, the relationships area in either tree or list views does not update automatically to show the alert. This is a bug.

changing in-use language codes

When a language code is changed or removed, any items that use the code remain unchanged. The only visible indication is that the language code in the item’s popup menu is disabled (allowing the user to select another code but not allowing the selection of an invalid code). This is quite deliberate and guards against unintended consequences. During export (XML, HTML or printing), language codes that are no longer valid are replaced with appropriate sentinel values (eg “unknown”).

using Cognatrix 1.5.1 & 1.6

The Cognatrix thesaurus file format is upwards compatible from earlier versions of Cognatrix (eg 1.5.1) to Cognatrix 1.6. The first time you use Cognatrix 1.6 to open a file saved under an earlier version, all items will default to a language-code of English. If you wish to work in both Cognatrix 1.5.1 and 1.6, you should keep separate copies of your thesaurus documents.

Both Cognatrix 1.5.2 and 1.6 include checks to help you avoid file-version problems. Cognatrix 1.5.2 will not open a document that has been saved by 1.6 (or later). The message that is displayed when this happens is “The document could not be opened.”

Partial Equivalence (PE) & Cognatrix 1.6.0b4 and later (BE/NE)

If you use Cognatrix 1.6.0b4 or later to open a thesarus document created by versions 1.6.0b1 through 1.6.0b3, any Partial Equivalence relationships will be silently discarded. This is because there is no way to automatically convert PE relationships to BE/NE pairs.

Licencing and registration

Cognatrix 1.6 respects the same licence as Cognatrix 1.5.1. If you do not have a licence, Cognatrix runs in demonstration mode.