Loading...
Difficulty
Your best score
N/A
The purpose of this problem is to identify and classify texts according to regional specificities or the standard form of the Romanian language. The dataset contains text fragments collected from various regions.
| Column | Description |
|---|---|
| ID | Unique identifier for each text fragment |
| text | The actual content of the phrase in Romanian |
| label | The dialect classification of the text (target column) |
Note: The label column (Dialect) is only available in the training data (train.csv). The three possible classes are: româna standard (standard Romanian), graiul moldovenesc (Moldavian dialect), and graiul bănățean (Banat dialect).
Some of the texts were extracted from the work "Cinci pâini" by Ion Creangă. Calculate the total number of occurrences of the word "pâni" (the archaic form of the word "pâini") in train.csv and test.csv.
In train.csv, calculate the average number of punctuation marks for texts in the graiul moldovenesc and graiul bănățean. Return the absolute value of the difference between the two values, rounded to 2 decimal places.
For each row in test.csv, calculate the number of diacritics in the text. The following characters are considered diacritics: ă, â, î, ș, ț (and their uppercase variants Ă, Â, Î, Ș, Ț).
Build a model capable of correctly classifying the texts from the test file into one of the 3 classes.
Evaluation is done by exact verification of the answer. Correct answer = maximum points, wrong answer = 0 points.
Evaluation is done via accuracy — the proportion of correct answers out of the total. The score is proportional: accuracy 1.0 = 10 points, accuracy 0.0 = 0 points.
Evaluation is done using the F1-Macro metric.
The submission file must be in CSV format, containing the following columns:
1.ID column in test.csv.subtaskID,datapointID,answer1,1,322,1,5.073,101,23,102,43,103,2...4,101,româna standard4,102,graiul moldovenesc4,103,graiul bănățean...Note: Solving this problem does not require the use of Transformer-type architectures (e.g., RoBERTa).