Identification based on typing patterns between programming and free text

Abstract

Identifying people based on their typing has been studied successfully in multiple different contexts. Previous research has shown that identification is possible based on writing predetermined texts such as typing passwords, free text such as essays, as well based on writing source code. In this work, we study typing pattern based identification when the text format and writing environment change. We replicate two earlier studies which suggested that typing profile identification works with programming data, and that it can be applied to a programming exam circumstances with decent results. Then, we examine how the identification accuracy changes when the user profiles are built using data from programming, and the identification is conducted on data from writing free text. Our results show that the identification accuracy is indeed high within the context of programming data, but drops when identifying essay typists based on typing profiles built from their programming data.

Programming Versus Natural Language

“…Context has previously been shown to have an effect on keystroke based identification accuracies [39] and predicting students' success based on keystrokes [13]. Comparing our results to those studies, we too found that context plays a part in our analysis.…”

“…For example, Villani et al [60] found that the type of keyboard can have an effect on the accuracy of identifying the person typing. Similarly, Peltola et al [39] found that when trying to identify students writing natural language based on typing profiles built from programming assignment keystroke data, accuracies were lower compared to when the context was solely programming assignments. Additionally, Leinonen et al [32] found that even when the context is the same -in their case, programming assignments -but the tasks are different (exam vs regular assignments), identification is not as accurate as when the context is more similar.…”

“…To our knowledge, however, there exists very little work that uses multiple sources of keystroke data. One of the few is [39], in which the authors sought to determine how the accuracy of identifying programmers based on typing changes across tasks.…”

Autenticação de alunos utilizando Dinâmica de Digitação e Redes Neurais Profundas em sistemas Juiz On-line

“…Para isso, os autores utilizaram o algoritmo de classificac ¸ão kNN e obtiveram cerca de 95% de precisão na identificac ¸ão dos alunos. O modelo de [Longi et al 2015] foi examinado em mais detalhes por [Peltola et al 2017], que descobriram que é possível identificar alunos em diferentes contextos, por exemplo durante a escrita de textos ou códigos de programac ¸ão. Em outro estudo, [Byun et al 2020] propuseram uma abordagem utilizando o algoritmo de classificac ¸ão Random Forest.…”

A Study of Keystroke Data in Two Contexts

“…during programming assignments. In a similar fashion, Peltola et al [26] studied how the type of text being written affects identification. They built typing profiles of students during programming assignments and examined how well students can be identified in a programming exam and when writing essays in natural language.…”