Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Improved the singularize method in inflect.py #220

Open
wants to merge 1 commit into
base: master
Choose a base branch
from

Conversation

TanyaaCJain
Copy link

Though 95% accuracy was previously achieved by measuring via CELEX English morphology word forms, the following changes have incremented the accuracy to 99%

  1. Added more words to the set singular_uninflected

  2. In the singularize method, changed the if-condition for the set singular_uninflected from
    if x.endswith(w): return word to if x == w or w == x + "s": return x
    because the former statement considered the words in the set to be word endings. Hence, it affected words with a prefix to the words in the set.
    The new condition checks if the word passed in the argument is present in the given list as it is or with a succeeding "s" and then returns the word's singular form from the list and not the word, which may be passed in a plural form.

  3. Added more words to the list singular_uncountable categorized via commenting such as abstract ideas and expressions, natural phenomena, general, etc for ease of reading and understanding

  4. Added more words to the list singular_ie and dictionaries singular_irregular

  5. Certain words which could be grouped via regex instead of adding in the above-mentioned lists and dictionaries were written in the form of regular expressions (regex) in the singular_rules.

  6. In singularize method, changed the if-condition for the dictionary singular_irregular from
    if w.endswith(x): to if x == w:
    because the former considered the word or key x in the dictionary to be an ending to the word passed as an argument to the singularize method. The latter condition checks whether the word w passed as argument is present in the dictionary by equating it to x. If True, it returns the singularized form of word w, that is, singular_irregular[x]

  7. Added more regex expressions to the list singular_rules to suit the singularization rules and improve the accuracy of the singularize method.

  8. Henceforth, this commit solves the following issues opened currently
    Issue - singularized on - earlier effect - current effect
    Singular form of words ending in 'our' and in 'lives' are incorrect #141 , issue singularizing "flour" #175 - flour - flmy - flour
    Singular form of words ending in 'our' and in 'lives' are incorrect #141 - colour - colmy - colour
    Singular form of words ending in 'our' and in 'lives' are incorrect #141 - your - ymy - your
    Singular form of words ending in 'our' and in 'lives' are incorrect #141 - olives - olife - olive
    issue singularizing "hummus" #176 - hummus - hummu - hummus

  9. The words added to sets singular_uninflected and singular_uncountable were also added to the lists in dictionary plural_categories["uninflected"] and plural_categories["uncountable"] for consistency.

It is to keep in mind that the 99% accuracy is reported after being tested from the corpora/test_en.py and is subject to the dataset of CELEX English morphology word forms only.

Though 95% accuracy was previously achieved on measuring via CELEX
English morphology word forms, the following changes have incremented
the accuracy to 99%

1. Added more words to the set singular_uninflected

2. In the singularize method, changed the if condition for the set
   singular_uninflected from
      if x.endswith(w): return word
   to
      if x == w or w == x + "s": return x
   because the former statement considered the words in the set to be
   word endings. Hence, it affected words with prefix to the words in
   the set.
   The new condition checks if the word passed in the argument is
   present in the given list as it is or with a succeeding "s" and then
   returns the word's singular form from the list and not the word,
   which may be passed in a plural form.

3. Added more words to the list singular_uncountable categorized via
   commenting such as abstract ideas and expressions, natural phenomena,
   general, etc for ease in reading and understanding

4. Added more words to the list singular_ie and dicts singular_irregular

5. Certain words which could be grouped via regex instead of adding in the
   above mentioned lists and dictionaries were written in the form of
   regular expressions (regex) in the singular_rules.

6. In singularize method, changed the if condition for the dictionary
   singular_irregular from
      if w.endswith(x):
   to
      if x == w:
   because the former considered the word or key x in the dict to be an
   ending to the word passed as an argument to the singularize method.
   The latter condition checks whether the word w passed as argument is
   present in the dict by equating it to x. If True, it returns the
   singularized form of word w, that is, singular_irregular[x]

7. Added more regex expressions to the list singular_rules to suit the
   singularization rules and improve accuracy for the singularize method

8. Henceworth, this commit solves the following issues opened currently
   Issue - singularized on - earlier effect - current effect
   141 , 175   - flour     - flmy           - flour
   141         - colour    - colmy          - colour
   141         - your      - ymy            - your
   141         - olives    - olife          - olive
   176         - hummus    - hummu          - hummus

   [141](clips#141)
   [175](clips#175)
   [176](clips#176)

9. The words added to sets singular_uninflected and singular_uncountable
   were also added to the lists in dict plural_categories["uninflected"]
   and plural_categories["uncountable"] for consistency.

It is to keep in mind that the 99% accuracy is reported after being
tested from the corpora/test_en.py and is subject to the dataset of CELEX
English morphology word forms only.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant