UIT The arctic university of Norway > Giellatekno


Fleksibel utdanning

Møte 7. april 2014

Til stede: Fran, Lene, Robert, Trond

Tidsplan under Fleksibel utdanning


Ryan now writes code for that, cf. the links for "Møter med Ryan" under Fleksibel utdanning.

Cf. the course goals for North Saami, here: http://giellatekno.uit.no/ped/fleksi/course/smecourse.html.

This then will be part of the course goal given to each student on her personal page. The program will then keep track of her work on each individual goal.

It is still a test, so do not make many goals. Just make some (relevant) goals, and test them out.

The X in the course goal list is a variable, a sign to Ryan to implement it e.g. with all books. The course goal list should be made by the teacher (Robert!) and implemented by Ryan.

Both teacher and student may log in, with different roles. Each student sees her teacher-given goals, and also sees the goals she has given to herself. She also sees the progress: How much she has worked, and what the results are (percent correct anwser).

The teacher does not see the self-given student goals, he only sees the goals he has given to all students, and for each and every student he will see the student progress. Even if the the course goals are defined for each course, the teacher will be able to edit them.

Later on, we may use the data for research, with anonymous student id-s , but identifying the progress of each individual student.


We now offer Google and Facebook, but should probably add vKontakte? The vKontakte would then be for Russian students wanting to learn Estonian. The students register as students of indivudual courses, but must be accepted by the teacher as a student of the course in question.


Look at Kven MorfaC (in the fkv and fkv_oahpa folders), and look at the http://oahpa.no/addlang/developerleksa.html and http://oahpa.no/addlang/developermorfa.html page for working with MorfaC (the latter to see the sets explanained).

Semantic sets

We have semantic variables, pointing to sets.

  • DRINK = tea, coffee
  • FOOD = soup, bread, ..
  • PEOPLE = boy, teacher
  • FAMILY = father, mother
  • MOVEMENT-verb = walk,

The lexicon contains all info for all lexemes, also the semantic class. One lemma can have several semantic classes.

The semantic classes are listed in meta/semantical_sets.xml. Those marked with an initial m are only for Morfa, and are not listed in this file.

Note there are class and subclass (semantic category).

In MorfaC not all these sets go toghether.

So, for Leksa, FOOD is an ok set. For MorphaC this is too coarse. We have FOOD_GROCERY, FOOD_DISH, and disthinguish between them. We lump them together in a class for Leksa, but use the subclass (semantic category) for MorfaC, so that you can buy FOOD_GROCERY, but order FOOD_DISH in the resaurant. You may also add subclasses to several classes.

The sets with m prefix are morfa sets. This just to remember that it does not need to be in the Leksa.


The frames are defined in the meta folder, e.g. meta/noun_questions.xml

Each frame has a unique frame. Variables written with capital letters, and defined.

  <q id="npar1">
      <text>Mitä SUBJ MAINV</text>
      <element id="SUBJ">
            <grammar tag="N+Sg+Nom"/>
        <sem class="PEOPLE"/>
      <element id="MAINV">
        <grammar tag="V+Act+Ind+Prs+Sg3"/>
      <text>SUBJ MAINV N-PAR</text>
      <element game='morfa' id="N-PAR" task="yes">
        <grammar tag="N+Sg+Par"/>
        <sem class="FOOD_DISH"/>

For each variable, there is an <element> explaining it.

semantical_sets.xml file

This file must exist, and it must have the same sets as the rest of the system.

Paradigm files

In meta we also need n_paradigm.xml, the tag list, listed exactly as the output of the fst. Note, no initial +, for historical reasons.

The file tags

Contains all tags, but again without the initial plus.


Not in use here, but needed by the system. Here we find the agreement.


We also have constants, like what, yesterday. We have declared all those in the file fillings.xml, in order to feed them to the mysql.

Restrictions in the lexicon files

When we generate the base, some lemmata do not have paradigms or have missing forms.

So, you have the tag gen_only="none", and with gen_only="none" the word is used only for Leksa. Also foreign names we may have in Leksa but we do not want to inflect it for Morfa.

We may have gen_only="Inf,Ind", which generates only infinitive and indicative, and no other forms of the paradigm.


We will use Zaliznjak. We do not need to implement this yet.

Progress plan

  • Draft of course goals by the end of this week? (meeting with Laura)
    • Take the sme goals and edit them
  • Simple MorfaC file
  • make rus/meta/semantical_sets.xml
  • fill in the n_rusnob.xml etc. files
  • Fran and Rob to do this soon.
  • Check by Lene and implementation by Heli, and then a demo on Friday.