Lynten / stanford-corenlp

Python wrapper for Stanford CoreNLP.
MIT License
919 stars 200 forks source link

Discrepancy in the result of dependency parsing with the corenlp server #61

Open ruihangdu opened 5 years ago

ruihangdu commented 5 years ago

Sentence: It's Lolita's chance of freedom," said Jared Goodman, PETA's director of animal law. "It's a huge step."

Server's output: screen shot 2018-09-20 at 9 41 46 am

Python corenlp's output: screen shot 2018-09-20 at 9 42 40 am

The server's output (and common sense) suggests that Goodman is the subject (nsubj) of the verb "said". However, the python version's output says that Goodman is the object of the verb "said". Could this discrepancy be fixed? Thank you.