Difference between revisions of "Extremaduran/Transducer"
(→Generator Analysis) |
|||
| (One intermediate revision by the same user not shown) | |||
| Line 6: | Line 6: | ||
==Generator Analysis== | ==Generator Analysis== | ||
| − | Our generator initially failed significantly more tests then our analyzer. After | + | Our generator initially failed significantly more tests then our analyzer. After writing four twol rules to account for multiple correct pronoun and determiner forms and adding support for several irregular verbs our generator now passes as many tests as our analyzer. Which is 73 out of 91 or 80%. |
| + | The addition of these rules increased our corpus coverage to 7.32%. | ||
| + | |||
==Code== | ==Code== | ||
Find our code at the following link: | Find our code at the following link: | ||
Latest revision as of 05:40, 8 March 2018
Transducer
Our transducer recognizes many verbs, determiners, and pronouns, as well as several determiner pronoun combinations.
Transducer Analysis
Our transducer currently boasts a 7.07% coverage rate on our corpus and passes 73 out of 91 tests.
Generator Analysis
Our generator initially failed significantly more tests then our analyzer. After writing four twol rules to account for multiple correct pronoun and determiner forms and adding support for several irregular verbs our generator now passes as many tests as our analyzer. Which is 73 out of 91 or 80%. The addition of these rules increased our corpus coverage to 7.32%.
Code
Find our code at the following link: https://github.swarthmore.edu/rbuckle1/ling073-ext.git
Notes
Adjective to adverb tests currently fail because we chose not to add those to our transducer and one verb form fails do to errors in the test file.
