vtt_3year_friends_ESF_annotation

Event Etructure Erame (ESF) annotation of the verbs in English scripts for the drama FRIENDS season1 episodes 01~05

Description

This project is about (semi-)automatic Event Structure Frame annotation of verbs in a given sentence by using ESF-annotated WordNet.

- Input: the list of text sentences

- Output: event structure frames for all verbs in the input text (json)

- Test data: English scripts of the American drama FRIENDS (season1 episodes 01~05)

Event Structure Frame Example

  sentence: Just leave my auro alone, okay?
  verb: leave
  Event Structure Frame Type: MAINTAIN
  Event Structure Frame:
         se1: pre-state: be(theme, state)
         se2: process: pred-ing(agent, theme, state)
         se3: post-state: be(theme, state)
         
  After argument mapping, you get the following:
         se1: pre-state: be(my_aura, alone)
         se2: process: leaving(my_auro, alone)
         se3: post-state: be(my_aura, alone)
  (pre-state: a presupposed state before the maintaining event, post-state: an entailed state after the maintaining event)

Requirements

- Python 3.0 or higher

- NLTK (Natural Language ToolKit) and NLTK Dataset

How to download and install NLTK: https://www.guru99.com/download-install-nltk.html

How To Use

Preparing input data

The input should be a list of sentences (csv).

Checking necessary data and program files

- main program: assign_esf.py

- pre-defined Event Structure Frame list: esf_list.py

- ESF-annotated WordNet: friends_esf_wn

- verb inflection forms: vInflection

How to run a main program

  python assign_esf.py "sentence"

Friends_verbs Event Structure Frame Annotation Result (json)

"Example: { "sec_num": "34", "sentence": "Just leave my aura alone, okay?", "verb_event_structure_frames": [ { "verb": "leave", "wn_synset": "leave.v.04", "esf_type": "MAINTAIN", "subevents": [ { "id": "se1", "se_type": "pre-state", "predicate": [ "be" ], "preposition": "", "arg_str": { "theme": "", "state": "" } }, { "id": "se1", "se_type": "process", "predicate": [ "refrain", "leave", "leave_alone", "forbear", "leave_behind" ], "preposition": "", "arg_str": { "agent": "", "theme": "", "state": "" } }, { "id": "se1", "se_type": "post-state", "predicate": [ "be" ], "preposition": "", "arg_str": { "theme": "", "state": "" } } ] } ] }"