Search for: All records

Abuse in cyber space is a problem requiring immediate attention. Unfortunately, despite advances in Natural Language Processing techniques, there are clear limitations in detecting instances of cyber abuse today. Challenges arising due to different languages that teens communicate with today, and usage of codes along with code mixing and code switching make the design of a comprehensive approach very hard. Existing NLP based approaches for detecting cyber abuse thus suffer from a high degree of false negatives and positives. In this paper, we investigate a new approach to detect instances of cyber abuse. Our approach is motivated by the premise that abusers tend to have unique facial expressions while engaging in an actual abuse episode, and if we are successful, such an approach will be language-agnostic. Here, using only four carefully identified facial features without any language processing, and realistic experiments with 15 users, our system proposed in this paper achieves 98% accuracy for same-user evaluation and up to 74% accuracy for cross-user evaluation in detecting instances of cyber abuse. 

Lying is a (practically) unavoidable component of our day to day interactions with other people, and it includes both oral and textual communications (e.g. text entered via smartphones). Detecting when a person is lying has important applications, especially with the ubiquity of messaging via smart-phones, coupled with rampant increases in (intentional) spread of mis-information today. In this paper, we design a technique to detect whether or not a person's textual inputs when typed via a smartphone indicate lying. To do so, first, we judiciously develop a smartphone based survey that guarantees any participant to provide a mix of true and false responses. While the participant is texting out responses to each question, the smartphone measures readings from its inbuilt inertial sensors, and then computes features like shaking, acceleration, tilt angle, typing speed etc. experienced by it. Subsequently, for each participant (47 in total), we glean the true and false responses using our own experiences with them, and also via informal discussions with each participant. By comparing the responses of each participant, along with the corresponding motion features computed by the smartphone, we implement several machine learning algorithms to detect when a participant is lying, and our accuracy is around 70% in the most stringent leave-one-out evaluation strategy. Later, utilizing findings of our analysis, we develop an architecture for real-time lie detection using smartphones. Yet another user evaluation of our lie detection system yields 84%-90% accuracy in detecting false responses.