ResultsIn this section, we will report the test of query generation algorithm based on one manually labelled entity the movie "Birdman". The metrics we will look at are Precision, Recall, F-Score and Accuracy.
The generated query has following keywords: Good keywords used: [u'boyhood', u'oscar', u'#birdmanmovie', 'watch', 'see', 'movie', 'film', 'trailer', u'congratulation', u'walking', ] Bad keywords used: [u'@birdman5star', u'sanchez', u'guess', u'code', u'hrithik', u'2018birdman', u'prevent', u'played', u'slid', u'heard', u'won', u'#sagawards', u'wayne', u'made', u'#golisten', u'check', u'deat', u'loved'] One More ResultAnother testing case is conducted on the entity "community", which is an American TV shown broadcasted on NBC.
The generated query has following keywords: Good keywords used: [u'proud', u'insight', u'season', u'follower', u'watched', u'show', u'check', u'working', u'#entrepreneurs', u'mosque', u'sweden', u'forward', u'enjoying', u'direction', u'posting', u'watching'] Bad keywords used: [u'photo', u'participate', u'hate', u'owler', u'tweet', u'sunday', u'blog', u'excited', u'facebook', u'joined', u'benefit', u'twitter', u'connect', u'god', u'bullshit', u'content', u'supporting', u'muslim'] |