Towards a triage model for online predator type prediction
Primary Investigator:
Julia (Taylor) Rayz
Kanishka Misra, Shih Feng Yang, Kathryn Seigfried-Spellar, Julia Taylor Rayz
Abstract
In an increasingly computer mediated world, it may be easier to start conversations with people online than it is in the real world, making it easy to build trust. This encourages people with malevolent intentions to take advantage of the trustful youngsters by means of various internet based messaging platforms. This poster presents results from applying classification methods on a corpus extracted from conversations between an online predator and a decoy. The Support Vector Machines algorithm is used to classify a predator as somebody who may or may not physically show up to meet their victim(decoy in this case). The features used in this study are word and character n-grams from the chat sessions. A nested resampling method is used to evaluate the performance of the classifier. The preliminary results indicate that the method can be used as a triage mechanism to help law enforcement.