Loading a phrase...

Entry Speed 0.00 Average 0.00
Error Rate 0.00% Average 0.00%

WebTEM: A Web Application to Record Text Entry Metrics

Ahmed Sabbir Arif

Launched on , last update pushed on

WebTEM is a Web application to record/gather text entry metrics. It works with almost all operating systems, browsers, and text entry techniques. Yet to verify compatibility, press the Check button and type a chunk of text to confirm that it detects all insertions and deletions. Press the Back button to return to the settings. To use the application, first enter a valid email address, study-specific information, and keyboard type. The application emails all logs to the email address provided and then deletes all data from the server. The study-specific settings are used for recordkeeping. Then select a phrase set, number of phrases in the session, performance metrics, and optional features. The most popular performance metrics and features are preselected for your convenience. Arif & Stuerzlinger, 2009 reviews and analyzes the most popular text entry metrics. International Metrics for non-Latin scripts are displayed when a non-Latin corpus is selected and calculated using the output stream convention proposed by Sarcar et al., 2015. The application displays the phrases in a randomized order. Press the Start button when ready to start the study. Make sure that the device has access to the Internet throughout the study.

The application generates one information (.INFO) and two tab-delimited files (.TSV) to record all settings, timestamped events, and performance metrics, respectively. Each row of the metrics log represents a phrase and each column a metric. The last row holds average values. Events are recorded as [time,text,event,duration,insertion,deletion,expected_input], where time is the number of milliseconds since January 1, 1970, text is the current state of the transcribed text, event is a user or system action (tap, autocorrection, etc.), duration is the time for the action in milliseconds, and insertion, deletion are character(s) entered and deleted by the action, respectively, and expected_input is the character the user presumably wanted to input. Read the paper below, visit the Update Log, and watch this video to know more about WebTEM. Please cite the following paper if you use WebTEM for your user studies.

Ahmed Sabbir Arif and Ali Mazalek. 2016. WebTEM: A Web application to record text entry metrics. In Proceedings of the 2016 ACM International Conference on Interactive Surfaces and Spaces (ISS '16). ACM, New York, NY, USA, 415-420. DOI: https://doi.org/10.1145/2992154.2996791

Feedback & Consultation/Collaboration—Contact me at to report bugs or request new features, metrics, and/or phrase sets for your user studies; if you are seeking consultation or want to collaborate on a research project evaluating a text entry technique(s).

Hide Introduction Hide Introduction
Log Files
Study Data Leave this field blank if you are creating a pre-populated form and want each participant to enter a predetermined, unique ID


Demographics
Female Male
Left Right Ambidextrous
Current Setting (Select All that Apply to You) DEVICE
Laptop
Tablet
Mobile
Smartwatch
Smart Glasses
TECHNIQUE
Non-Qwerty
Gesture Typing
SETTING
Outdoor
Indoor
LOCATION
Work
Home
School
Meeting
MOBILITY
Seated
Walking
Running
Driving
Riding
Flying
Commuting
Keyboard Type Virtual Keyboard Select this option if the keyboard type is unknown or predictive metrics are desirable for a physical keyboard
Physical Keyboard
Phrase Sets (Scroll down for International Sets)
The quick brown fox jumps over the lazy dog
500 Short English Phrases MacKenzie & Soukoreff, 2003
36 Very Short English Phrases Subset of MacKenzie & Soukoreff, 2003
200 Memorable English Phrases Vertanen & Kristensson, 2011
900 Mixed-Case Alphanumeric Text & Special Symbols Arif et al., 2014
500 Children's English Phrases Kano et al., 2006
50 Emo English Phrases
111 Bengali Phrases Without Numbers & Punctuations Arif & Fardeen, 2016
300 Marathi Phrases Without Numbers & Punctuations Dalvi et al., 2016
Custom Phrase Set (One Phrase per Line) Non-Latin/English Phrases
Metrics Phrase Number
Presented Text (PT)
Transcribed Text (TT)
Presented Text Length (|PT|)
Transcribed Text Length (|TT|)
Input-Stream Length (|IS|)
 Words per Minute (WPM)
Characters per Second (CPS)
Error Rate (ER)
Minimum String Distance Error Rate (MSD ER)
Keystrokes per Character (KSPC)
Corrected Error Rate (CER)
Total Error Rate (TER)
Input Time (IT) IT signifies the time (in seconds) users took to enter a phrase
Visual Scan Time (VST) VST signifies the time (in milliseconds) users took to visually scan a recently completed phrase, mainly to proofread, before submitting it
Cursor Control Count (CCC) CCC is the total number of times users repositioned the cursor using the arrow keys, the mouse, direct touch, or a digital pen to correct errors or to edit text in the process of inputting a phrase
Backspace Count (BC) BC is the total number of backspaces per phrase
Corrective Action Count (CAC) CAC is the total number of corrective actions, particularly backspaces and direct cursor control, per phrase
Corrective Action Rate (CAR) CAR is the ratio (%) of the total number of corrective actions to the total number of actions per phrase
Total Corrective Action Time (T-CAT) T-CAT is the total time (in seconds) the user took to perform the corrective actions per phrase
Corrective Action Time Ratio (CAT-R) CAT-R is the ratio (%) of the total time the user took to perform the corrective actions to the total time s/he took to enter a phrase
Prediction Rate (PR) PR is the ratio (%) of the total number of characters automatically entered by the predictive system and the total number of characters entered in a text entry episode. It does not account for incorrect predictions, since they are difficult to identify
Levenshtein Substitution (Substitution Error) Levenshtein Substitution is the total number of substitutions needed to change the transcribed text to the presented text, usually caused by substitution errors.
Levenshtein Insertion (Omission Error) Levenshtein Insertion is the total number of insertions needed to change the transcribed text to the presented text, usually caused by omission errors.
Levenshtein Deletion (Insertion Error) Levenshtein Deletion is the total number of deletions needed to change the transcribed text to the presented text, usually caused by insertion errors.
Presented Text Output Stream Length (|OSP|)
Transcribed Text Output Stream Length (|OST|)
 International Metrics Characters per Second (CPSint)
Error Rate (ERint)
Minimum String Distance Error Rate (MSD ERint)
Keystrokes per Character (KSPCint)
Corrected Error Rate (CERint)
Total Error Rate (TERint)
Additional Options Show an "Enter" button under the input area
Show asterisks instead of actual characters while typing
Show asterisks instead of actual characters for the current word while typing
Disable predictive features
Ignore letter case and extra spaces in metrics calculation
Ignore extra spaces at the end of transcribed phrases in error rate metrics calculation
Ignore extra emojis in error rate metrics calculation
Present all phrases in lowercase
Present all phrases without special characters
 Display all/unique phrases from the set in random order (number of phrases should not exceed the total phrases in the set)
Display the number of phrases entered
Display performance summary
Hide presented text when users start typing
Force error-free submissions (includes auditory feedback)
Keep equal spacing between presented text and input area, input area and keyboard
Display a "Start" button to override the automatic detection of the start of text entry
Add border around the presented text
Hide the virtual keyboard on "Enter" to force users to tap on the input area to see the keyboard
Link to a Pre-Populated Form

Fill in the form and click on the Generate button to create a weblink to a pre-populated form. With pre-populated forms, participants can start participating in a study immediately. However, select from the options below, if you require your participants to enter unique IDs (shared with them ahead of time), demographics, and their current position/settings, which can be beneficial in "in-the-wild" studies. You can distribute the link via social network platforms, emails, text messages, etc. with your participants. Use the Short URL button to shorten the link for your/your participants' convenience.

Participant ID
Demographics
Current Setting

Disclaimer—The application is distributed in the hope that it will be useful, "AS IS" without warranty of any kind, either expressed or implied, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose. The entire risk as to the quality and performance of the program is with you. Under no circumstances shall the authors have any liability hereunder for any loss or damage caused directly or indirectly from the use of this application.