layout: page title: POS tagging 5. Exercises ——————————-

Back: Korean POS tagging

POS analysis in another language

Choose a language that you are interested in, other than English.

Select an NLP pipeline that supports your chosen language and identify:

  • the pipeline or model used;
  • the UPOS tag set;
  • the language-specific POS tag set, if available.

Select your texts

Prepare or select multiple texts in your chosen language.

Your texts may come from:

  • the provided sample corpus;
  • an existing corpus;
  • texts that you collect yourself.

Use at least five text files. Briefly report the source, genre, and language of the texts.

Conduct the POS analysis

Process all texts using the selected pipeline.

For each text:

  1. generate word-level or token-level annotations;
  2. report the lemma, UPOS tag, and language-specific POS tag;
  3. exclude punctuation from the POS frequency counts;
  4. count each UPOS category;
  5. count each language-specific POS tag;
  6. create summary tables comparing the texts.

Submit the following files

Submit one ZIP file containing:

  • analysis_description.txt Language, text source, pipeline, model, UPOS scheme, and language-specific tag set.

  • token_annotations.csv Filename, token or word, lemma, UPOS, and language-specific POS tag.

  • upos_counts.csv UPOS frequencies for each text.

  • language_specific_pos_counts.csv Language-specific POS-tag frequencies for each text.

  • all_texts_upos_counts.csv A wide-format table with one text per row and one UPOS category per column.

  • all_texts_language_specific_pos_counts.csv A wide-format table with one text per row and one language-specific tag per column.

Name the compressed file:

lastname_firstname_pos_analysis.zip

Back to Module 7