Please enable JavaScript.

Coggle requires JavaScript to display documents.

Identifying Sarcasm in Twitter: A Closer Look (Information (Methodology…

- - - - But human judges other than the tweets’ authors, achieve low levels of accuracy when trying to classify sarcastic tweets; we therefore argue that using the tweets labeled by their authors using hashtag produces a better quality gold standard.
      - To build our corpus of sarcastic (S), positive (P) and negative (N) tweets, we relied on the annotations that tweeters assign to their own tweets using hashtags, the best judge of whether a tweet is intended to be sarcastic is the author
      - We used a Twitter API to collect tweets that include hashtags that express sarcasm (#sarcasm, #sarcastic), direct positive sentiment (e.g., #happy, #joy, #lucky), and direct negative sentiment (e.g., #sadness, #angry, #frustrated), respectively.
      - To address the concern of Davidov et al. (2010) that tweets with #hashtags are noisy, we automatically filtered all tweets where the hashtags of interest were not located at the very end of the message.
      - We then performed a manual review of the filtered tweets to double check that the remaining end hashtags were not part of the message
      - We thus eliminated messages about sarcasm such as “I really love #sarcasm” and kept only messages that express sarcasm, such as “lol thanks. I can always count on you for comfort :) #sarcasm”
      - Lexical and Pragmatic Features
        
        Feature Ranking
        
        In all of the tasks, negative emotion (Negemo), positive emotion (Posemo), negation (Negate), emoticons (Smiley, Frown), auxiliary verbs (AuxVb), and punctuation marks are in the top 10 features.
        
        To measure the impact of features on discriminating among the three categories, we used two standard measures: presence and frequency of the factors in each tweet
        
        We did a 3- way comparison of Sarcastic (S), Positive (P), and Negative (N) messages (S-P-N); as well as 2-way comparisons of i) Sarcastic and Non-Sarcastic (SNS); ii) Sarcastic and Positive (S-P) and Sarcastic and Negative (S-N)
        
        We also observe indications of a possible dependence among factors that could differentiate sarcasm from both positive and negative tweets:
        
        We ran a χ 2 test to identify the features that were most useful in discriminating categories. Table 1 shows the top 10 features based on presence of all dictionary-based lexical factors plus the pragmatic factors. We refer to this set of features as LIWC+ .
        
        Sarcastic tweets tend to have positive emotion words like positive tweets do (Posemo is a significant feature in S-N but not in S-P), while they use more negation words like negative tweets do (Negate is an important feature for S-P)
        
        Lexical Factors
        
        We used two kinds of lexical features – unigrams and dictionary-based
        
        Feature derived from Pennebaker et al.’s LIWC (2007) dictionary,
        
        Consists of a set of 64 word categories grouped into four general classes: Linguistic Processes (LP) (e.g., adverbs, pronouns), Psychological Processes (PP) (e.g., positive and negative emotions), Personal Concerns (PC) (e.g, work, achievement), and Spoken Categories (SC) (e.g., assent, nonfluencies);
        
        Also ii) WordNet Affect (WNA) (Strapparava and Valitutti, 2004); and iii) list of interjections (e.g., ah, oh, yeah)1 , and punctuations (e.g., !, ?).
        
        Pragmatic Factors
        
        We used three pragmatic features: i) positive emoticons such as smileys; ii) negative emoticons such as frowning faces; and iii) ToUser, which marks if a tweets is a reply to another tweet
        
        We merged all of the lists into a single dictionary.
      - We applied automatic filtering to remove retweets, duplicates, quotes, spam, tweets written in languages other than English, and tweets with URLs.