AffectEventKB is a knowledge base of events developed by the NLP group at University of
Utah. The events in the knowledge base were extracted from personal stories which were
identified from web blogs. Each event is represented using a frame-like tuple which
consists of 4 fields: Agent, Predicate, Theme, PrepPhrase(PP). Each event is associated with
a polarity label (highest score), and a distribution of scores over three polarities:
positive, negative, and neutral.

The released data directory contains:
1. affective event lexicon
2. gold standard event polarity annotations
3. a script to choose vary sizes lexicons


======== Affective Event Lexicon =======

./affevent.csv

  A csv file with each line representing an event. There are 9 columns.
  Column 1-4: "Agent", "Predicate", "Theme" and "PrepositionPhrase". All words are lemmatized.
  Column 5: The predicted event polarity "pos(positive)" or "neg(negative)" or "neu(neutral)".
  Column 6-8: The polarity score over three classes: positive, negative and neutral;
              Column 5 shows the polarity label with the highest score.
  Column 9: The frequency of the event in our data.

  The following shows an example event ``@I@ break @my@ leg''.

        *********** Example *********************
        @I@,break,@my@,leg,neg,0.9,0.09,0.01,100
        
        Column 1-4: @I@,break,@my@,leg
        Column 5: neg
        Column 6-8: 0.9,0.09,0.01
        Column 9: 100
        *****************************************



======== Event Polarity Annotations =======

./dev.csv
./test.csv

  There are 1490 events with human annotated polarity labels.
  The development set has 490 events; the test set has 1000 events.
  Both are csv files. Each line represents an event. There are 5 columns.
  Column 1-4: "Agent", "Predicate", "Theme" and "PrepositionPhrase". All words are lemmatized.
  Column 5: The annotated event polarity "pos(positive)" or "neg(negative)" or "neu(neutral)".

  The following shows an example event ``@I@ break @my@ leg''.

        *********** Example *********************
        @I@,break,@my@,leg,neg

        Column 1-4: @I@,break,@my@,leg
        Column 5: neg
        *****************************************



====== Pronoun Normalization ======

  Please note that, pronouns in the events are normalized using the following mapping: 
	   @I@           <--   I, me, myself
	   @my@          <--   my, mine
	   @we@          <--   we, us, ourselves
	   @our@         <--   our, ours
	   @you@         <--   you, yourself, yourselves
	   @your@        <--   your, yours
	   @he@          <--   he, him, himself
	   @his@         <--   his
	   @she@         <--   she, her, herself
	   @her@         <--   hers
	   @they@        <--   they, them, themselves
	   @their@       <--   their, theirs



======== The "main.py" script ========

In this directory, we also provide a script to output varying sizes lexicons by
constraints. You can specify polarity labels, frequency threshold, confidence
score threshold, input/output file. We also provide a more human-readable
output format. Please run "python reformat.py -h" for more information.

