Code Reusability Tutorial

Build up the solution for a new field

The first step is to create a language model for the specific subject.
The "Build Language Model" file contains an example of how to build a vocabulary for 
solar observations, but in the same way it is possible to train a new language model for
any subject. It connects to Zenodo to get as many papers as possible on the subject.

It is necesary to build up a hierarchical onthology for any specific subject, that
will be added in the "Predefined Labels" file. For a minimal number of changes in the source
code, the hierarchy should follow the rules in the file: start from "themes" related words,
then "domains" and "information"

The algorithm can be found in the 'Supervised Classifier' file. Firstly, the user needs to build
up a database on the subject. In this case, we used Google Scholar API to get abstracts, as short
texts on the subject and we keep them in files. For each entry, we saved the ids, the titles, the
abstracts and the urls.

We follow the steps of the algorithms, described in the documentation and we change only the name
of the variables.