Prepare a Corpus (Corpus Builder)

N.B.: The pictures shown in this section have been obtained by using a previous version of T-LAB. In T-LAB 10 this tool includes two additional buttons: a) one, named Text Screening, which becomes enabled when the corpus size is up to 20 MB; b) the other which allows the user to immediately proceed with the import of selected textual materials (see the below picture).

This software tool is intended to simplify and speed up any transformation of documents and textual materials into a corpus file ready to be processed by T-LAB.

More specifically, such a tool allows the following operations:

1. Automatically import various types of files;
2. Edit and tag them by using categorical variables;
3. Save the result as a corpus file ready to be imported by T-LAB;
4. Check and modify any corpus file which corresponds to the T-LAB format.

While the way that files are imported (see '1' above) varies according to their format, all the other operations follow the same logic.

Below is a short description of how to import the various files.

A - Importing files in tabular or spreadsheet format (CSV, .SAV, .JSON, .XML, .XLS, XLSX, .MDB, .ACCDB).

A single file which includes up to 30,000 records can be imported by the 'Open Table' option or by a simple drag and drop method (N.B.: When text for each line doesn't exceed 2.000 characters, up to 99.999 records can be imported).

The fields/columns of such a file can contain the following data:

- Categorical Variables (one for each column, up to a 50 columns);
- Texts to be analysed (only one column);;
- IDnumbers, i.e. identifiers of subjects (e.g. in the case of answers to open-ended questions) or of context units in which the corpus to be imported is subdivided.

N.B.: While the presence of Categorical Variables and IDnumbers is optional, the presence of at least one column containing the texts to be analysed is mandatory.

When importing a .CSV file, the corresponding delimiter must be selected (see below).

When importing Excel or Access files, only one table can be selected (see below).

B - Importing document files of various formats

Up to 30,000 documents can be imported, either one by one or by multiple selection, through the 'Gather your Texts' option (see below). Three methods are available:.

The first method ('Add your Documents') allows the automatic importation of .TXT, .DOC, .DOCX, .PDF and .RTF files.

The second method ('Add EmptyRecord') allows the user to copy/paste any type of text (see below).

The third method ('Import Text from URL') allows downloading HTML files from Internet, as well as editing their content before the importation (see below). .

C - Importing a corpus file already encoded according to the T-LAB specifications.

The use of the 'Open Corpus' option is advisable in the following three cases:

1 - the user intends to modify the structure of a corpus already encoded (e.g. add further texts by means of methods explained in the previous 'B' section, modify the labels of variables and values, etc.);
2 - the user intends to check/fix errors of his manual coding that had been possibly done without the aid of the Corpus Builder module;
3 - the user intends to import a corpus file with a 'raw' coding, that is a corpus the sections of which (i.e. documents or records) are preceded by a coding line with four asterisks ('**** ')., just followed by a blank space and a return key (see below).

.

In all the above cases (1,2,3) it is sufficient to select a single file by means of the 'Open Corpus' option (see below) or use the drag and drop method.

Operations that follow any file importation

At the end of the phase with which the files have been imported in Corpus Builder, either when the user is 'not' interested in the use of variables or when the encoding operations have already been carried out, he may proceed with the 'Check / Generate' option and afterwards with the exportation of the corpus to be imported in T-LAB.

When the corpus is encoded it should be recalled that in all three types of importation mentioned in the preceding sections of this document ('A', 'B', 'C') data are displayed in various columns, the headers of which can be the following:

- Variable, i.e. categorical variables, the use of which is requested when analysing the characteristics and the reciprocal relationships of various corpus subsets;
- IDnumber, i.e. identifiers of cases/records, the use of which is optional;
- My Texts, i.e. the texts to be analysed, the use of which is mandatory and is allowed in a single column only;
- Exclude, the use of which indicates that data in the corresponding column(s) must not be saved by the Corpus Builder module.

In all cases it must be remembered that:
- each record can be selected or deselected (see below '1'),
- IDnumbers can be automatically added (see below '2'),
- each variable name can be edited and changed (see below '3');
- each variable value can be edited and changed (see below '4')
- each 'My Text' field can be edited and changed too (see below '5').

Further information:

- the number of columns with categorical variables must not exceed 50;
- each variable can have a maximum of 150 values;
- the IDnumber values, if used, must be progressive starting from 1 (e.g. 1, 2, 3, etc.);
- each label, both for variables and values, must not exceed the length of 25 alphanumeric characters (at least 2) and must not be interrupted by blank spaces;
- when doing any operation, all detected errors are visualized in the bottom-left window (see below).

Use of the Variable Manager tool

The 'Variable Manager' tool allows the user to build, edit, load, save and change any coding scheme, even from a different corpus.
Each coding scheme includes the list of variables and that of their values (see below) .

In order to add variables from a different corpus or from a previously scheme saved the 'Load Variable Scheme' option must be selected (see above '1'). Otherwise, to manually add variables and their values, the sequence of the above '2' and '3' options must be used.

Any value assigned to individual records must be added manually (see below) and in a single work session, this is because when saving coding schemes the values of each record are not recorded. Consequently, when the user is dealing with a corpus that includes a considerable number of records and / or his job requires more than one session, it is recommended to proceed as follows:

1 - import the amount of files / records that can be encoded in a single work session;
2 - save one's work as a corpus (see the 'Save' option in the Corpus Builder menu);
3 - then, in the subsequent session, re-import the corpus previously saved (see above, point '2 '), add further records / files to encode and continue.

When the basic operations have been carried out (i.e., two or more texts have been gathered), by clicking the 'Check/Generate' button the user can verify the correctness of his work and export (A) or save (B) a corpus ready to be imported by T-LAB.

In the first case (A - see below) Corpus Builder creates a new folder under the directory '..\My Documents\T-LAB PLUS\" and automatically starts the importation procedure.
N.B.: In this case the new folder has the same name of the corpus file.

In the second case (B - see below) the user is enabled to save his corpus in whatever folder he wishes and aftwerwards he has to use the 'Import a corpus' option of T-LAB.
N.B.: In this case it is recommended that a new folder be created which should contain only the corpus to be imported.

.