2. Getting started with DALIA

2.1. How to install DALIA.

To install the program, you can download a precompiled binary, if there is one for your platform, or compile from the sources.

For version 2.1, precompiled binaries are available for FreeDOS, Linux (i386) and Win.

2.1.1. How to install a precompiled binary.

To install a precompiled binary, just extract the binary .zip file to the folder you want. Run DALIA from its current directory. To run, DALIA needs that some configuration files be present in the same directory in which DALIA is. These files are alredy present in the package you downloaded.

For LINUX, you must remember to create a new folder called "temp" in the same directory, before you run DALIA. If the LINUX version shouldn't run, try changing the rights and permissions of the files.

2.1.2. How to compile from the sources.

To compile DALIA's source code, you will need, first of all, to download and install the FreePASCAL http://www.freepascal.org compiler for your platform (it has recently been released for MAC too). To compile DALIA:

o Get the FreePASCAL compiler and install it following the instructions;

o Change the file OS_SPEC.pas according to the needs of your operating system (for Linux, Win and DOS you only have to uncomment the appropriate linex, and comment out those of the other operating systems);

o Compile the source code;

o For DALIA to work, you need to download the "grammars.zip" file from DALIA's distribution, unfold it, and move the compiled binary to that folder.

2.2. The configuration files.

To work, DALIA needs the following configuration files:

o ``lingvoj.txt'', a list of the available languages. Each language must have a .lex and a .pho file;

o ``arities.txt'', a list of the arity prefixes;

o ``ug_verb.txt, ug_nuon.txt, ug_pr.txt'', three files which containthe predicate hierarchy.

You can customize the configuration files to suit your needs. These files have the following structure:

2.2.1. lingvoj.txt

Each line of this file contains an entry for a language. Each entry begins with the name of the language (there are no restrictions on this), then the language code, between square brackets, for example:

English [eng]

Quechua (Bolivian) [qub]

Codes should be at most three characters long, and are case-sensitive.

You can use a complex code: after the language code, you put a colon, then the dialect code. For example,

Quechua (Bolivian) [qu:bol]

Quechua (Ecuadorian) [qu:eq]

See``Dialects'' for more information about how DALIA can handle various dialects of a same language.

2.2.2. arities.txt

This file must contain the arity labels 0, 1 and 2, to which you can add the prefixes you choose as aliases for predicate arities. Each predicate used in semantic representations must have an arity, that is, a label specifying the arguments each predicate can take.

Predicates are written

arity\category:predicate

where category can be omitted.

The file arities.txt must contain three lines. The first lists the aliases for 0-arity, the second and the third, respectively, for 1- and 2- arity predicates. There can be other (comment) lines, which must begin with the character %. Example:

%ARITY 0

e d 0

%ARITY 1

E D p 1

%ARITY 2

r 2

2.2.3. ug_verb.txt, ug_noun.txt, ug_prep.txt

These files contain the hierarchy for functional heads, respectively, in the verbal, nominal and adpositional projections. Hierarchy is from the top down (higher elements are written after lower ones), in increasing order (the last element is the highest in the hierarchy, the first is the lowest). Each line contain just one entry, that for the relevant category for the hierarchy. See the single files in the distribution.

The hierarchy of functional heads is needed to expand the LF of a sentence for translation ito another language.