The software projects terms of serious game models (in UML) into a set of user defined taxonomies (pref. pedagogical or ludic) to build four metrics: Precision, Clarity, Overlap and Specialization. The software includes algorithms to scrap thesauri from the api, to match model terms to taxonomy terms and to generate CSV, PNG or TXT files to visualize the data.
The software was developed using spyder and anaconda. The python libraries used are :
- nltk.corpus
- urllib3
- colorama (needed for manual exceptions treatment only)
- datetime
- xml.etree.ElementTree
- requests
You will also need to download the following github files/repositories :
The calculation and synthetisation of metrics are done in 3 steps :
- Running scrap_thesaurus_data to generate local TXT files containing the entire thesaurus structures. The download takes several hours with a computer scientist PC (Intel Core Ultra 7 155H (16 Cœurs HT, 2.5-4.8 GHz Turbo, 24Mo cache).
- Running matching to generate CSV files and PNG images,
- (Not required) Running data_presentation to compute all the CSV data in one file.
The test_matching file is useful to check that the functions of matching are correct. If needed, functions to separate a taxonomy into its branches are available in matching.
The TXT files are written using python language. It is just a way to store locally the thesauri, avoiding hours of download each time you want to run the algorithms.
The CSV files present the mean values of the four metrics used to compare the twelve serious game models chosen in our comparison with eleven educational taxonomies (precision, clarity, specialization, and overlap). The generated file all_csv organizes the values with serious game models and taxonomies in columns and metrics in rows, grouped by taxonomies. The precision, clarity, specialization, and overlap files present the values with serious game models in columns and taxonomies in rows.
The PNG images illustrates the precision and specialization of serious game models for each taxonomy. These visualizations are presented as box plots.
The method implemented by the software has been published in the proceedings of CSEDU 2026. The full data associated to this article are available at https://recherche.data.gouv.fr
This work was funded by the ANR project TALE4GDA (ANR-23-CE38-0001) : https://tale4gda.wp.imt.fr/.
The number into parenthesis is the number of elements of the models:
- (Marne et al., 2012) (9),
- (Ildephonce et al., 2018) (12)
- (Elborji and Khaldi, 2014) (20)
- (Pellon et al.,2020) (21)
- (De Troyer et al., 2017) (32)
- (Djaouti et al., 2011) (33)
- (Mariais et al., 2012) (43)
- (Prismatik) (48)
- (Arnab et al., 2015) (79)
- (Amory, 2007) (83)
- (Avila-Pesantez et al., 2017) (95)
- (Carvalho et al., 2015) (473)
The list of the taxonomies used:
- Councill of Europe; REDINED (n.d.). European Education Thesaurus
- CEDEFOP European Training Thesauru
- Education Services Australia Schools Online Thesauru
- National Centre for Vocational Education Research VOCED Thesaurus.
- Intrallect Learning activities.
- International Standard Classification of Education: fields of education and training
- Tesaurus for Education Systems in Europe Eurydice
- European Schoolnet LRE thesaurus.
- ACL, ACLearn
- USA. Institute of Education Science, ERIC Thesurus
- UNESCO IBE Education Thesaurus English (2026-03-11)