Laboratoire ICAR » Plateforme TXM

A. More synthetic display - words & score - whole lexicon¶

Current specificity results are displayed in the following columns (VOEUX/text@loc:word):

Unit       F  f loc1  sp loc1  f loc2  sp loc2 ...
.       2461     848     17,2     325    -24,3 ...
Une       35      26      8,2       4     -1,0 ...
La       135      65      7,0       8     -6,0 ...
aussi    117      56      6,1      13     -2,4 ...
emploi    60      34      6,0       3     -3,3 ...

To explore specific words for two different parts,
the user needs to successively sort each 'score' column
of the parts considered to get access to the most/less
specific words.

An initial specificity results display could be more
synthetic by providing immediately the most/less specific
words for each part, higher/lower than the banality threshold.

Here are possible displays for most/less specific words:

u. loc1  sp loc1  u. loc2  sp loc2 ...
      .     17,2        ;     14,4 ...
    Une      8,2        ,     11,2 ...
     La      7,0      qu'     10,6 ...
  aussi      6,1       en     10,2 ...
 emploi      6,0  Algérie      9,3 ...

or even more synthetic

  loc1      loc2 ...
     .         ; ...
   Une         , ...
    La       qu' ...
 aussi        en ...
emploi   Algérie ...

With this display, the user begins to read specificity results globaly at first: which units are specific to each part.

Means could be provided to navigate to the other classical view of specificity results for more detailed information.

A. Solution¶

Add a new "TXM/User/Specificity/Specificity display" preference with values:

• full -> current specificity display with unit, f and sp
• synthetic (default value) -> for synthetic display

Add a toggle button in the specificity editor to switch between the two display types.

B. More synthetic display - score only - index lexicon¶

To be even more synthetic, one can only display the specificity scores higher/lower than the banality threshold.

This display mode would be useful for specificities based on an index lexicon (not a full lexicon).

Here is a possible display:

unit       sp loc1  sp loc2  sp loc3 ...
peuple        -4,7      4,2          ...
emploi         7,6     -5,3          ...
avenir         5,9                   ...
liberté                          5,8 ...
confiance                            ...

B. Solution¶

• add a new 'TXM/User/Specificity/Filter specificity scores' parameter, with 'true' as default value
• when FilterSpecificityScores is true, don't display specificity score under the banality threshold in the results table

Add a toggle button in the specificity editor to switch between the two display modes.

• add a new "TXM/User/Specificity/Display frequencies" parameter, with 'true' as default value
• when DisplaySpecificityFrequencies is false, don't display the frequencies columns

Add a toggle button in the specificity editor to switch between the two display modes.

C. Even more synthetic display - pattern only - index lexicon¶

To be even more synthetic, one can only display the pattern of scores higher/lower than the banality threshold.

The pattern could be read horizontaly or verticaly.

This display mode would be useful for specificities based on an index lexicon (not a full lexicon).

Here is a possible display:

unit       sp loc1  sp loc2  sp loc3 ...
peuple           -        +          ...
emploi           +        -          ...
avenir           +                   ...
liberté                            + ...
confiance                            ...

or

unit       sp loc1  sp loc2  sp loc3 ...
peuple           -        +        . ...
emploi           +        -        . ...
avenir           +        .        . ...
liberté          .        .        + ...
confiance        .        .        . ...

or

unit       sp loc1  sp loc2  sp loc3 ...
peuple           -        +        b ...
emploi           +        -        b ...
avenir           +        b        b ...
liberté          b        b        + ...
confiance        b        b        b ...

C. Solution¶

Add a "pattern" value to the "TXM/User/Specificity/Specificity display" preference of Solution A.

See also:

• Le vocabulaire des confédérations syndicales ouvrières : une analyse des spécificités, Alain Bergounioux, Michel F. Launay, Josette Lefềvre, René Mouriaux, Jean-Pierre Sueur, Mots Year 1981 Volume 2 Issue 1 pp. 139-156, http://www.persee.fr/doc/mots_0243-6450_1981_num_2_1_1025
• Le vocabulaire spécifique des motions Mitterrand, Rocard et CERES au congrès de Metz (1979), Simone Bonnafous, Mots Year 1981 Volume 3 Issue 1 pp. 79-94, http://www.persee.fr/doc/mots_0243-6450_1981_num_3_1_1040