Deception Detection via Language-Action Cues and Histograms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing use of computer-mediated communications poses risks of online deception, including social engineering, spam, phishing, identity theft, and fraud, necessitating effective methods to detect deceptive content and protect users.
Innovation Solution
A system and method that parses digital text to identify language-action cues, uses a machine learning-based classifying algorithm trained with deceptive and truthful data to analyze and produce probabilities of deceptive content, and displays a histogram indicating these probabilities over time, combining with other deception-detecting techniques like facial expression and speech analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If computer-mediated communications are used to enable digital text, audio, and video communication, then user communication productivity and accessibility are improved, but users become vulnerable to online deception, social engineering, spam, phishing, identity theft, and fraud
Solution Approach 1:
The patent introduces an intermediary deception detection system that analyzes digital text communications to identify deceptive content. This intermediary layer processes communications between users, using machine learning algorithms to detect language-action cues and generate deception probabilities, thereby protecting users from online deception while preserving the benefits of computer-mediated communication
Solution Approach 2:
The system implements feedback by providing real-time or near-real-time deception detection results to users. The histogram display shows frequencies of deception probabilities over time, giving users feedback about the trustworthiness of communications they encounter, enabling them to make informed decisions about engaging with potentially deceptive content
2Measurement precision
If machine learning algorithms are trained with large datasets of deceptive and truthful communications to improve detection accuracy, then measurement precision of deception detection is improved, but device complexity and data processing requirements increase
Solution Approach 1:
The patent segments the deception detection task into distinct components: parsing digital text to identify language-action cues, analyzing these cues with machine learning algorithms, and displaying results as histograms. This segmentation allows the system to process complex deception detection through manageable stages, reducing overall system complexity while maintaining detection accuracy
Data Source
AI summary
Provided are methods and systems for detecting deception in computer-mediated communications that can include (i) parsing digital text to identify processable language-action cues in the digital text, (ii) automatically analyzing, using a computing device and using a machine learning-based classifying algorithm, the processable language-action cues to produce respective probabilities of the digital text including the deceptive content, where the machine learning-based classifying algorithm is trained with training data including (A) training language-action cues from digital data indicating response information known to be deceptive and digital data indicating response information known to be truthful and (B) respective predictor weights associated with the training language-action cues, and (iii) displaying a histogram on a user display device, where the histogram indicates, over a period of time, frequencies of the respective probabilities of the digital text including the deceptive content. Other methods, systems, and computer-readable media are also disclosed.