Frame-Based Detection of Figurative Language in Tweets [Application Notes]

摘要: This paper analyzes the problem of figurative language detection on social media, with a focus use semantic features for identifying irony and sarcasm. Framester, novel resource that acts as hub between FrameNet, WordNet, VerbNet, BabelNet, DBpedia, Yago, DOLCE-Zero others, has been used to extract from text. These are enrich representations tweets event information using frames word senses in addition lexical units. The data set experimentation purposes contains taken different corpora including both (containing sarcasm) non-figurative language. Two major tasks were performed: (i) detecting dataset containing tweets, (ii) classifying A 10-fold cross-validation experiment shows obtained accuracy increases significantly when such linguistic units, indicating they may be important clues approach was developed top Apache Spark so it is easily scalable much higher volumes data, allowing real-time analysis.