The underlying lexicon model is the Eagles model (Expert Advisory Group to Language Engineering Standards). More precisely only the subset of objects that was interesting to us was implemented.
According to the windows used it is possible to manage a monolingual lexicon or a multilingual lexicon. A monolingual lexicon contains words (called morphological units) and meanings (called semantic units). So a meaning is an object belongs to a certain language. The translation from a language to another does not occur at the word level but at the meaning level. Each meaning is connected to a concept that is not specific to a language: a concept is considered as a pivot.
TagEdit permits to manage words and meanings of a particular language and, without having to exit from the workstation to change the language. It should be noted that at every moment there is a current language.
The window dedicated to multilingual operations permits to manage translations. It is of course possible to manage a bilingual lexicon, it is just a particular case of multilingual editing.
Here is a brief snapshot of the functionalities:
The lexicon can also be accessed thru an API that is not described here.
Interactrive Editing [top of page]
At the session start, the main window is displayed, with the tab "Interactive Editing" selected as a default value:
The left column is dedicated to written forms of words. These latter are typed with the keyboard (for a true interactive use) or loaded from a file. In this latter case, the written forms come from a batch processing done by the user or another person that prepared the work for the user. Instead of giving a character string, the user can give a regular expression: something like ".*tion" in order to get all the words that end with "tion". The right column is to type an optional comment concerning the work of the word. It is something like: "done" or "to check".
The fact that the string is typed or loaded has not incidence on the rest of the session.
We the user click on "Edit" the written form is look up in the lexicon. Depending on the preferences (see screen) the search is computed from a lemmatized form, an inflected form (an underlying morphological parser is called) or an object identifier.
The screen then shows the candidates computed from the search:
The top part is present when at least a candidate is found: this is to modify existing words. The bottom part is always present even if candidates are found: this is to add new words.
If edition of a new word is choosen, the following screen is showed. This is the window for morphology editing of the word and the link to the meanings of the word.
If modification of a words is choosen, selection of a words is needed then click on "Modify". The following screen will be displayed. This is the edition of one meaning.
Batch management [top of page]
This is to launch subprograms that import, export and verify.
Counting screen [top of page]
The goal is to sum up the content of variouns lexicons in order to have a quantified view of it.
The screen does not allow any modification.
It is possible to record an array in the file, for instance in order to include it in a report.
Preference screen [top of page]
The regular lexicographic mode permits to create a words, its meanings and its translations: that's it.
The expert mode permits to view and modify the hidden information or the built information. For instance in the expert mode, it is possible to view and modify administrative attributes: last modification date. It is possible to edit linguistics constants like inflection objects or the syntactic properties of a language.
The screen for preferences is as follows:
