##What is TypeShift?

TypeShift is a series of modules written over the past few years in collaboration between CUNY, NYIT and Louisiana Tech. These modules are used to extract and modify feature sets pulled from a proprietary test data set of keystroke dynamics. These feature sets can then be analyzed for the purposes of user authentication.

##The Data

No data is contained in this repository due to intellectual property liscensing. However, a description of the data is:

The typing data was collected from 486 Louisiana Tech University students (hereinafter referred to as "subjects") in two sessions, 6 months apart. Subjects received unique IDs identifying them across sessions, and we collected a number of self-reported demographics for each subject.

The subjects were 41% female to 59% male. 82% are L1 English speakers, while 17% are non-native English speakers. Finally, 88% of subjects report being right-hand dominant, while 9% report being left-handed.

The mean typing rate was 168.9 intraword keystrokes per minute with a standard deviation of 50.04.

In each session, subjects were presented with a set of 12 prompts to respond to (from a set of 36). The order that the prompts were presented was randomized. There is no overlap of prompts across sessions.

Examples of prompts include:

• List the recent movies you've seen or books you've read. When did you see or read them? What were they about?

• How would you design a class if you were the teacher? What subject would you teach? How would you structure your course?

Each subject was required to type at least 300 characters in response to each prompt, but was given unlimited time. The average response contains 448 characters and 87 words. Our system recorded the timing associated with the subjects' keystrokes, and cursor position.

##src Folder The src folder contains all of the existing modules that have been written. These modules are divided among the two institutions where they were written. In the subfolders of cuny and nyit you will find the various modules. The modules and their intended purpose is outlined below.

##CUNY Modules

Demographics Cog Load 2 Way Cog Load 3 Way Cog Load 6 Way Gender Handedness Primary Language Query All Subject ID

Demographics Extractors Demographic Extractor Handedness

Keyboard Canonical Finger Character Type Transition Finger Hand Hold Finger Hand Speed Finger Hand Transition Speed Finger Hold Finger Row Transition Speed Finger Speed Finger Transition Speed Hand Finger To Row Transition Speed Hand Hold Hand Speed Hand to Hand Row Speed Hand to Row Speed Hand Transition Speed Lag Times Row Finger Transition Speed Row Hand Hold Row Hand Speed Row Hand to Hand Speed Rown Hand to Row Hand Speed Row Hold Row Speed Row to Hand Finger Transition Speed Row to Hand Speed Row Transition Speed Session Typing Rate

Keytouch Key Touch Context Cum KH2KI

Key Touch Context KHKI Trigraph Contextualizes each unigraph hold and digraph interval in trigraph context

Key Touch Context Ngram

Key Touch Context Uni Hold DiInt

Key Touch Fusion Ngram

Key Touch Fusion Sep KHKI Digraph Contextualizes each unigraph hold in digraph context

Key Touch Fusion Uni Hold DiInt Implements a specific version of ngram fusion. It takes a unigraph key hold and digraph key interval

Key Touch Hold Generates full permutation of two UpperCase alpha combinations and places them into a TreeSet which ensures no duplicates and automatically sorts them.

Key Touch Represents an entire key cycle from press to release for slurs. KeyTouch list is ordered by first key press.

Key Touch Ngram Highly modular code to allow for selection of ngram length, as well as whether to include interval pauses, keyholds or both.

Key Touch Sub Ngram

Key Touch Trigraph

Key Touch Unigram

Lex EXTRACTORS Bigram Extractor Create IDF Map Create Spell Check Map Create Syntax Model Create TFIDF Map Create User Pause Metrics Map Create Word Frequency Map Extract All Answers POS Extractor Sentence Detector Spacebar Punctuation Metrics TF IDF Extract TF IDF Token Extender Tokenize Visual Char Stream

MODULES Character Type Consonant Frequency Consonant Timing Count Sent Word Char Edit Metrics Function Word Metrics In Revision Keystroke Count Lex Events Btw Pause Lexical Diversity Linear Regression Pearsons Coefficient POS Metrics POS Pause Punctuation Pause Spelling Metrics TF IDF Metrics Token Extract Typing Bursts Typing Rate Viz Char Stream Metrics

Mwe Mwe Extractor Mwe Timing Demog Mwe Timing Token Extended

Ngram Model Bigram Bigram Model Fourgram Fourgram Model Generate Model From COCA Language Model Generator Model Generator Exception Trigram Trigram Model Unigram Token List Generator

Pause Between Pauses Duration Average Between Pauses Duration Between Pauses Keystroke Count Average Key Up Down Key Up Down Stats KSE Pause Bursts Pause Count Pause Count Stats Pause Duration Pause Duration Stats Pase Vs Typing Rate Run Pause

Predictability Consonant Timing Predictability Digraph Duration Pred Digraph Expanded Duration Pred Digraph Longest Duration Pred Digraph Z Hand Transition Predictability Key Hold Predictability Key Hold Pred Key Interval Key Stroke Predictability Kt Duration Category Pred Pk Duration Pred Mwe Timing Pred Demog Mwe Timing Predictability Pennebaker Pred POS Predictability Predictability Test Predictability Touch Zone Predictability Trigraph Duration Trigraph Duration Pred Trigraph Z Word Plus KS Predictability Word Predictability

UID Word Token Surprisal

Revision Revision CSV Revision Revision Ngram Fusion Revision Ngram ReVisualize Test Parse Revision Unigraph Hold Revision

Util Aggregator Arff Manipulator Availability Prune Calculate Expected Values Convert Test Vectors Correlation Tester Keystroke Correlation Tester Word Csv to Arff Expected Value Calculator Generate Classifier File Keystroke Map Keystroke Pointer Make Symmetric Test Vectors Measure Word Timing Merge Arffs Process Arff Experiment Process Weka Pipeline Test Extraction Module Test Vector Creator Test Vector Exception Test Vector Processor Trigram Counter Vector To Weka Converter Weka Pipeline Exception Weka Pipeline

WordVar WordVariationCSV

##NYIT Modules

AppWindow AppWindow AppWindowScriptAG AppWindowScript

Ballchair getThStats

Events Event Iterator Event List Event Parser Generic Event

KEYSTROKE RESEARCH DocState T Event Iterator T Event T Key Event T Mouse Event T Parser T Time Event

PHASE 6 7 Key Stroke Iterator Key Stroke Key Stroke Parser

Extractor Answer By Order Comparator Data Node Dat Selector Db Config Extraction Module Feature Extractor Feature Test Vector Shutdown Module

Lex/modules Character Type Erase Pattern Familiarity

Module Ancillary Data Test Atomic Slurs Cog Load Cognitive Load Example Digraph Duration Example Module 2 Feature Survey Feature Word Count Four Graph Duration Key Hold Key Interval Keystroke Duration Serializer Key Stroke Survey Key To Text Test Latency Stats Pause After Word PP Burst P P Burst PP Burst Metrics PP Burst Test P PR Burst PR2 Burst PR Burst PR P Burst PR PR Burst RP2 Burst RP Burst RR Burst Slur Density Slur In Word Split Trigraph Duration Time Analysis Trigraph Duration Word Count

Post Processing PP Burst Postprocess PR Burst Postprocess Remove Features Remove Invalid RP Burst Postprocess RR Burst Postprocess

Tools Ancillary Data Interface Availability Analyzer Average Value Analyzer Convert Generics Custom Pipeline Pipeline Modules List Raw Data Fetcher Segment Answer Slur Capture Test Data Text Data Import