Crystalace - Sarcasm Detection Engine for Enhanced Sentiment Analysis in Text
01 Jan 2021
Detecting sarcasm is among one of the toughest problems in AI. In computational linguistics and NLP, sarcasm detection is receiving increasing research interest. However, while recent studies recognized the linkage between sarcasm and sentiment
and have proposed various techniques for detecting sarcasm, none directly and systematically studied the impact of sarcasm detection on sentiment analysis.
- Analyse a wide range of written/spoken natural language text input such as tweets, Facebook posts, comments, news headlines and articles, or speech transcripts
- Produce highly accurate prediction outputs in terms of sarcasm vs non-sarcasm classification result as well as the confidence score
- Ground with explainable and theoretically sound affective AI research behind the system, with industry-strength software robustness
The Science Behind
The underlying predictive algorithm underneath Crystalace sarcasm detection engine has novel sociolinguistics-inspired, psychologically meaningful and explainable features. Crystalace has high detection accuracy with its F1-score approximates 0.66 tested
with human-annotated ground truth data.
Crystalace’s accurate sarcasm detection has been found to be useful in enhanced sentiment analysis over multiple use cases. It reduces misinterpretation of opinions and attitudes, in particular in contexts when there are high likelihoods of people
expressing negative meaning while using salient, seemingly positive language cues (e.g., comments on public services and policies, feedback on extremely negative experiences).
Test out our interactive demo here: Crystalace
For more info or collaboration opportunities, please write to firstname.lastname@example.org.