Abstract
Numerous false information come up to be the primary threat for regular communication and cooperation, which always have similar expressions and diffusion patterns with regular information. To recognize deceptive information and alleviate their threat against business and economy, a number of research approaches for deceptive information identification have been proposed. However, these approaches suffer from their rough feature extraction processes, and thus cannot distinguish deceptive information from others. We introduce an adaptive slide-window based feature extraction, which captures semantic features and eliminates the trivial parts from them with adaptive slide windows on sentence elements, in order to facilitate accurate semantic structure representation of various texts. In addition, we propose a deep deceptive information identification model based on the proposed feature extraction scheme. Experiments on three real-world datasets demonstrate that the proposed deep deceptive information identification model can distinguish deceptive information from regular information accurately by extracting the significant features instead of trivial common expressions.
Highlights
Augmented reality (AR) terminals and other advanced mobile devices leverage emerging technologies to support novel on-line applications in recent years
FEATURE SCORING AND EXTRACTION To find out the features that have significant contributions to deceptive information identification, we study the semantic structure of textural information, and leverage slide windows to achieve a feature scoring scheme for searching important features on different semantic scales
ADAPTIVE SLIDE WINDOW BASED FEATURE REFINEMENT To eliminate the trivial words from the extracted features, we extensively explore the semantic structure of textural information and apply the observation to guide adaptive slide windows for refining feature extraction
Summary
Augmented reality (AR) terminals and other advanced mobile devices leverage emerging technologies to support novel on-line applications in recent years. To make deceptive information more deceptive, malicious users only make few changes on the real information, while sharing many similar expressions and other primary explicit features with the honest information For this example, the deceptive information differs from the honest report only by several words. The similar trivial text expressions will involve significant noises in deceptive information identification. To eliminate the trivial expressions from each feature as much as possible, we propose an advanced adaptive feature extraction method which can incorporate comprehensive semantic structure representation of texts into a convolutional neural network (CNN). An enhanced deceptive information identification model, adaptive window feature extraction based convolutional neural network model with Gated recurrent unit (AWFE-CGN), is obtained. Based on an observation about the probabilistic difference of semantic correlations between deceptive information and regular information, we propose an advanced adaptive window mechanism to eliminate trivial words from features.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
