To participate in the shared task, you will build a system that can predict typological properties of languages, given a handful of observed features. Training examples and development examples will be provided. All submitted systems will be compared on a held-out test set.
To obtain the final results, run python scripts/score.py [TSVFILE] [more TSVFILES]. Note that the script runs on cleaned input files which may not be the file you submitted. Generated plots are in results_plots.ods, they do not interact with the python script.
The model will receive the language code, name, latitude, longitude, genus, family, country code, and feature names as inputs and will be required to fill values for those requested features.
Input:
mhi Marathi 19.0 76.0 Indic Indo-European IN order_of_subject,_object,_and_verb=? | number_of_genders=?
jpn Japanese 37.0 140.0 Japanese Japanese JP case_syncretism=? | order_of_adjective_and_noun=?
The expected output is:
mhi Marathi 19.0 76.0 Indic Indo-European IN order_of_subject,_object,_and_verb= SOV | number_of_genders=three
jpn Japanese 37.0 140.0 Japanese Japanese JP case_syncretism=no_case_marking | order_of_adjective_and_noun=demonstrative-Noun
The model will have access to typology features across a set of languages. These features are derived from the WALS database. For the purpose of this shared task, we will provide a subset of languages/features as shown below:
tur Turkish 39.0 35.0 Turkic Altaic TR case_syncretism=no_syncretism | order_of_subject,_object,_and_verb= SOV | number_of_genders=none | definite_articles=no_definite_but_indefinite_article
jpn Japanese 37.0 140.0 Japanese Japanese JP order_of_subject,_object,_and_verb= SOV | prefixing_vs_suffixing_in_inflectional_morphology=strongly_suffixing
Column 1: Language ID
Column 2: Language name
Column 3: Latitude
Column 4: Longitude
Column 5: Genus
Column 6: Family
Column 7: Country Codes
Column 8: It contains the feature-value pairs for each language, where features are separated by ‘|’