A crowd-sourced study on a deep-learning based mobile keyboard app
-
- STATUS
- Not Recruiting
Summary
A huge amount of textual content is generated by the Internet users. Every day there are 2 million comments posted to Reddit, 500 million Tweets, 3.5 billion Google search queries, at least 100 billion instant messages, and 200 billion emails sent and received. A significant proportion of these texts are typed in from mobile devices such as smartphones and tablets.
To make mobile users type faster, the research community and industry have developed numerous Input Method Applications (IMAs). An important component inside modern IMAs is Next-word Prediction. An accurate prediction of the next word a user is going to input reduces the number of clicks as well as the probability of typing errors. This feature is especially appealing on mobile devices where keyboards are small and fingers are big.
Under the above context, we have developed an IMA called DeepType. A key feature of DeepType is to use personalized deep learning to perform next-word prediction. For example, given the same prefix ?every Friday night I usually go to the ___?, Alice tends to input ?gym? next, while Bob is likely to input ?club.? DeepType performs deep personalization so that every individual device uses a different, personalized prediction model, which fully adapts to the preference and context of the user.
In this study, we let real users install and use DeepType on their mobile devices. We will collect various performance data such as the accuracy of our next-word prediction model, the prediction latency, and the training latency. Participants are not exposed to any risks other than those of normal smartphone users. A potential concern we are aware of is the privacy issue. To protect participants? privacy, in this study we do NOT collect any texts typed by them. Also, the DeepType app itself does not upload any user-typed data to any external Internet server or cloud. A complete list of the types of data we collect is listed in the next question.
We will use the collected data to assess and improve DeepType?s performance ?in the wild?.
Description
A huge amount of textual content is generated by the Internet users. Every day there are 2 million comments posted to Reddit, 500 million Tweets, 3.5 billion Google search queries, at least 100 billion instant messages, and 200 billion emails sent and received. A significant proportion of these texts are typed in from mobile devices such as smartphones and tablets.
To make mobile users type faster, the research community and industry have developed numerous Input Method Applications (IMAs). An important component inside modern IMAs is Next-word Prediction. An accurate prediction of the next word a user is going to input reduces the number of clicks as well as the probability of typing errors. This feature is especially appealing on mobile devices where keyboards are small and fingers are big.
Under the above context, we have developed an IMA called DeepType. A key feature of DeepType is to use personalized deep learning to perform next-word prediction. For example, given the same prefix ?every Friday night I usually go to the ___?, Alice tends to input ?gym? next, while Bob is likely to input ?club.? DeepType performs deep personalization so that every individual device uses a different, personalized prediction model, which fully adapts to the preference and context of the user.
In this study, we let real users install and use DeepType on their mobile devices. We will collect various performance data such as the accuracy of our next-word prediction model, the prediction latency, and the training latency. Participants are not exposed to any risks other than those of normal smartphone users. A potential concern we are aware of is the privacy issue. To protect participants? privacy, in this study we do NOT collect any texts typed by them. Also, the DeepType app itself does not upload any user-typed data to any external Internet server or cloud. A complete list of the types of data we collect is listed in the next question.
We will use the collected data to assess and improve DeepType?s performance ?in the wild?.
Details
| Condition | healthy |
|---|---|
| Age | 100years or below |
| Clinical Study Identifier | TX8982 |
| Last Modified on | 19 February 2024 |
Similar trials to consider
Not finding what you're looking for?
Sign up as a volunteer to stay informed
Every year hundreds of thousands of volunteers step forward to participate in research. Sign up as a volunteer and receive email notifications when clinical trials are posted in the medical category of interest to you.
Sign up as volunteerStudy Definition
WikipediaAdd a private note
- Select a piece of text.
- Add notes visible only to you.
- Send it to people through a passcode protected link.