MANDINGA
Word Sense Induction Algorithm
Version: June 20, 2026
For now, this version is
only available for Spanish.
Examples of polysemous words in Spnish
agente,
bajista,
cólera,
papel,
pintura,
planta,
ratón,
viral
.
This script accepts a single-word noun in Spanish,
extracts a sample of concordances from a general corpus,
and applies a graph-based clustering algorithm that, it is here claimed,
results in a reflection from the noun's senses, in case it has more than one.
This program was created in May 2008 and used to find the different
references of initialisms/acronyms, as part of
Nazar (2010) PhD thesis
. It was also used then with a diachronic corpus
in order to detect semantic neologisms,
and it was presented at the CINEO Conference in 2008 (Nazar & Vidal, 2010)
.
A more detailed description of this
Word Sense Induction algorithm is available (Nazar, 2013).
References
Nazar, R. (2010). A Quantitative Approach to Concept Analysis. PhD Thesis. IULA, Universitat Pompeu Fabra.
http://www.tdx.cat/TDX-0117111-085812
Nazar, R. (2013). Word Sense Discrimination Using Statistic Analysis of Texts. Barcelona Research Art Creation, vol. 1, no. 1.
https://www.hipatiapress.com/hpjournals/index.php/brac/en/article/view/608
Nazar, R.; Vidal, V. (2008). Aproximación cuantitativa a la neología. En
Mª. Teresa Cabré, Ona Domènech, Rosa Estopà, Judit Freixa y Mercè Lorente (eds.)
Actes del I Congrés Internacional de neologia de les llengües romàniques, CD-ROM.
Barcelona: IULA.
http://www.tecling.com/nazar/CINEO_Nazar_Vidal.pdf