Progress in Automatic Flirtation Detection

“Detecting human social meaning is a difficult task for automatic conversational understanding systems.” – explain a research team [pictured] based at Stanford University, who have investigated the viabilities of an automatic flirtation detector.


“Our flirtation detection system uses prosodic, dialogue, and lexical features to detect a speaker’s intent to flirt with up to 71.5% accuracy […]”

The high figures suggest that their system can predict flirtation intention even better than people can – humans can only manage to predict flirtation intention with 56.2% accuracy – it’s claimed.

See: Ranganath, Rajesh, Dan Jurafsky and Daniel A. McFarland, 2009. “It’s Not You, It’s Me: Detecting Flirting and Its Misperception in Speed-Dates.” Empirical Methods on Natural Language Processing (EMNLP, 2009, Singapore). Session 3A (Theatre): ‘Discourse and Dialogue’.

Also see: Development of the Flirting Styles Inventory (2010)

Note: Previous academic work on automatic detection of human social intentions via speech analysis has focussed on annoyance, anger, sadness, or boredom [Ang et al., 2002; Lee and Narayanan, 2002; Liscombe et al., 2003], speaker characteristics such as charisma [Rosenberg and Hirschberg, 2005], personality features like extroversion or agreeability [Mairesse et al., 2007; Mairesse and Walker, 2008], speaker depression or stress [Rude et al., 2004; Pennebaker and Lay, 2002; Cohn et al., 2004], and dating willingness or liking [Madan et al., 2005; Pentland, 2005] ).