Instruction:
First Step: Please select the keywords from the following sentence.
Second Step: Rate each keyword base on importance.
For example:
Given sentence: My watch fell in the water. We can select "watch", "fell", "water" as keywords. Then we enter the following data in to the textarea below:
watch fell water
Notice: "Blank" to separate the different keyword

Instructions切换成中文

From existing research, we learned that writing as a cognitive process begins with the brain proposing an abstract idea of what to write (pictures, individual words, or phrases/clauses at most). The abstract idea is usually semantically complete, and our brain will transcribe it into the final form (correct spelling and grammar, appropriate tone, etc.).

In this study, we will focus on understanding what kind of information will be abstracted by our brain first. The process is simulated by presenting a complete sentence to you first; after you get the complete semantic meaning of the sentence, please identify the words that will occur in your mind if you want to convey the same information. We will refer to them as keywords.

For example, if you want to write a short sentence like "Let's have a quick drink tonight after the movie." Keywords that are very likely coming to your mind would be "drink", "tonight", "movie", and "quick". We want you to identify these keywords by typing them in the textbox below, and also rate them based on how important they are according to the semantic meaning of the whole sentence. 1 means the least important keyword in the sentence (e.g., "quick"), 5 means the most important (e.g., "drink", and/or "tonight"). Note that you can give same scores to multiple keywords if they are same important.

Progress

Phrase Count

Use Enter for Rate/Next

Settings

Stop test after phrases

Ignore Case

Tutorial

Welcome to TextTest++, a WEB platform for text entry test. The output of the test data can be seamlessly plugged in the throughput calculation project here . We build our platform purely with JS, thus you won’t worry about network-delay when testing. After loading the webpage, you’re good to go!

P.s. We strongly recommend using Chrome browser, as other browsers might yield unexpected results because of their different implementations of the web api.

Before Test

Import Phraseset

By clicking “Settings”, you can find the option Upload Custom Phrase Set. We use the phrase set from Phrase sets for evaluating text entry techniques as default set. Make sure your phrase set uses the following format (one phrase for one line):

phrase 1
phrase 2
…
phrase n

After choosing the phrase file, click upload to import the phrases.

Randomize Phrases

The default order of phrases is the same order as the phrase file. To randomize the test order, you could click the "Shuffle whole set" button.

If you check the box of "Stop test after", and modify the number, then the test will end after the number of phrases are completed.

Other Settings

Ignore Case - check Ignore Case in settings to ignore case between comparison of Present string and Transcribed string
Shortcut for Next - Check Use Enter for Next to enable change to the next phrase by hitting Enter key (the option is next to button)

During Test

During the test, the left corner of the input text will show the count of finished phrases. If the user entered all phrases, it will turn to red showing "task done!"

Restart last phrase

If the user want to re-enter the last phrase (for example, the user accidentally hit Next before finish the current phrase), he/she could click last button to restart the last phrase. However, one can only restart one phrase at one time. All actions of the last phrase before restart won’t be logged in the file.

Restart whole test

You can just refresh the webpage to start a new test or restart a test. Remember to set the phrase amount, seed, upload new phrase set (if necessary) and repeat other settings again.

Log section

You will notice that when the input text changes, the gray-background area below will change its content. The content shows the information of the text change using the approximation algorithm of T-sequence model.

After Test

Congratulations! You’ve finished a test. Now you might want to download the log files and further analyze them. Remember, always hit Analysis before download the file.

Analyze data

Click Analysis button to show the graph of different error rates of the experiment. It is always required before download any file.

A statistic graph will show the different error rates (TER/UER/CER) of the typing process. A typical graph will look like this:

The graph is especially useful when you want to "clean out" some accidental data. for example, if one accidentally hit Next before he is finishing the phrase, you might find the phrase using the graph, because the UER (Uncorrected Error Rate) will be very high of the phrase.

Download Data

You can set the custom file name or leave the filename section empty, which will generate a file named “TextTest”. Select the type of file and download. You can download all types.

Type	Contents
json	the log file of action information for each phrase
csv	the metrics file for each phrase, analyzed from the json file

(for detailed explanation, please go to “Reference of csv file variable names”)

After analyze or download the experiment, you can always continue the test if you don’t refresh the page. The log file will continue to log the experiment.

Import an existing json file

You could analyze an existing json file in other experiments as long as the format is the same. After successfully uploading, the error rate graph will appear, and you can download the CSV result of the file.

Variable Name in CSV	Explanation
Trial	Index of the trial
Seconds	Total time spent to type the phrase
correct_time	Total correct time (delete + substitute) using approximation algorithm
entry_time	Total entry time (insert + substitute)
Tlen	Transcription string (T) length
Plen	Presented string (P) length
TCC	Total Changed Characters during the typing process of the trial TCC = Tlen + 2 × IF
IF	IF (Incorrect Fixed) class, contains all deleted characters in the T-sequence
INF	INF (Incorrect Not Fixed) class, Minimum string distance (MSD) between P and T
C	C (Correct) class, contains correct characters in T
WPM	Word per minute
TOTPM	Total Word Change per minute
AC	Action count, the number of actions taken during the text entry process
DAC	Delete action count
IAC	Insert action count
SAC	Substitute action count
UER	Uncorrected Error Rate UER = INF / (C + INF + IF)
CER	Corrected Error Rate CER = IF / (C + INF + IF)
TER	Total Error Rate TER = (INF + IF) / (C + INF + IF)
CPA	Character per action, indicates the average number of characters changed per action CPA = TCC / AC
CPC	Character change per correct action CPA = TCC / AC
CPE	Character change per entry action CPE = (\|T\| + IF) / EAC
AE	Action Efficiency, the average number of characters one action can change in a given time period AE = TCC / Total time
CE	Correction action Efficiency, indicating the “text-correcting speed” of actions CE = IF / Correction time
EE	Entry action Efficiency, indicating the “text-entering speed” of actions EE = (Tlen + F) / Entry time
IFc	Correct characters in IF
IFe	Error characters in IF

JSON Format

A typical trial will be recorded in the following format:

    [
        {
            "Transcribe": [
                {
                    "Text": "m",
                    "TimeStamp": 1517175820224
                },
                {
                    "Text": "my",
                    "TimeStamp": 1517175820417
                }
            ],
            "Action": [
                [
                    "insert",
                    0,
                    1
                ],
                [
                    "insert",
                    1,
                    1
                ]
            ],
            "Trial": 0,
            "Present": "my watch fell in the water",
            "IF": 0,
            "INF": 24,
            "C": 2,
            "CER": "0.000",
            "UER": "0.923",
            "TER": "0.923"
        }
    ]

Transcribe contains the sequence of the transcription texts, each has Text and TimeStamp property
Action contains the sequence of the actions corresponding to each transcription text, each has Operation (Insert/Delete/Substitute) and range (for insert, it's (start index, inserted count)/ for delete, it's (start index, end index)/ for substitute, it's (start index, end index)) property
Trial means the current trial index
Present is the presented string
IF/INF/C/CER/UER/TER is the corresponding metric

Main Page

How to use?

Reference of the JSON and CSV file variable names

Keywords Ranking

Rate keywords below!