LAC: Language and the Computer: Project 3

In-Class On-Line Open-Book Programming Challenge

This assignment constitutes 30% of your final grade for LAC.

Comparing and describing wordnets

Given a pair of wordnets, compare them. Given a single wordnet describe it or illustrate it.

Please start with these wordnets:

• Tokyo University of Foreign Studies Basic Vocabulary for English
  • Contains words and example sentences for basic vocabulary
  • Web page is here: https://fcbond.github.io/tufs/
  • Wordnet web page is here: https://malindo.aa-ken.jp/TUFSOpenLgResources.html
  • Vocab Search page is here [in Japanese]: https://www.coelang.tufs.ac.jp/mt/index_v.html
• Arasaac for English
  • Contains words and links to pictures
  • Definition is a space-separated list of picture ids
  • Web page is here: https://arasaac.org/
  • Wordnet web page is here: https://github.com/bond-lab/wn-araasac/
• Then test with similar wordnets, you can chose the languages
  • TUFS: ['as', 'de', 'en', 'es', 'fr', 'id', 'ja', 'km', 'ko', 'lo', 'mn', 'my', 'pt', 'ru', 'th', 'tl', 'tr', 'ur', 'vi', 'zsm']
  • ARASAAC: ['an', 'ar', 'bg', 'ca', 'cs', 'de', 'en', 'es', 'et', 'eu', 'fa', 'fr', 'gl', 'he', 'hr', 'hu', 'it', 'ko', 'lt', 'mk', 'nb', 'nl', 'pl', 'pt', 'ro', 'ru', 'sk', 'sr', 'zh']
• Core synsets are a semi-automatically compiled list of 5000 "core" word senses in Princeton WordNet (approximately the 5000 most frequently used word senses). We have mapped them to ilis
  • wn-core-ili.tab
• Basic synsets are a compiled list of 578 "basic" synsets from the Tokyo University of Foreign Studies Basic Vocabulary for Japanese. We have mapped them to ilis
  • wn-basic-ili.tab

Single wordnet description

Write a function (or functions) that take a wordnet lexicon id as input, assuming the wordnet has already been read.

1. How many synsets, senses and words are there?
2. How many synsets, senses and words are there for each pos (part of speech)?
3. Show these as e.g. pi charts
   Make the size of the pi chart proportional to the actual size
   E.g if one wn has 100,000 synsets and one has 5,000, the first pi chart should be 20 times bigger
4. Show three examples of each pos

Compare wordnets

1. How many synsets, senses and words are there only in one, in both or only in the other?
2. Show these as e.g. bar charts --- left for one wordnet, right for the other
3. Show three examples of words in one, both or other for each pos

Compare wordnet to curated list

1. Try this with the core and basic lists
2. How many synsets are in the wordnet from the curated concept list
3. See the coverage over the semantic fields (lexicographer files)
4. Graph this and give examples
5. Compare the two lists

Illustrate a basic list

1. Go through the basic list, and show a picture for each word
   • Easiest to do by concepts
   • Also show synonyms and examples, if any
   • You can get a definition from another wordnet (like en-omw:1.4)
2. Try to break things into smaller groups (pos, semantic field)
3. If the word has not picture, experiment with a hypernym or hyponym
4. Not how many words are missing pictures, and list them
5. You can show images in google colab like this:

   from IPython.display import display, Markdown
   
   ara_id=2910
   
   display(Markdown((f"""**Computer**
    
   <img width='48' src='https://static.arasaac.org/pictograms/{ara_id}/{ara_id}_300.png'> *from Arasaac*""")))
         

Deliverables

• Per person:
  • 1-2 page paper in any format, finally given as pdf, describing what you have done, and what you think could be done given more time.
    Include title, author, date e.g. LAC Project 3: Comparing wordnets
    Francis Bond
    bond@ieee.org
    2024-12-18
    The file should be named name_surname_matric.pdf
    e.g. francis_bond_007.pdf
    you do not need full citations, this is just a short summary
    don't include the program as an appendix (just share a link)
  • program
    The program should be in a google colab
    Include the URL in the paper
• Upload them to moodle
  If you don't want me to show off your code/output/paper, please let me know within a month

Pedagogical Goals

• Consolidate skills in a novel task under realistic conditions
  You can do interesting tasks!
• Demonstrate an ability to break a task into small chunks
  Should be able to consider and evaluate multiple approaches
• Demonstrate an ability to use python to solve tasks
  choosing the right approach, writing and debugging
• Demonstrate an ability to cooperate to solve a problem
  good communication, delivering (partial) results in time

---

Project Three for LAC: Language and the Computer Francis Bond.