When a user initiates text selection within an application, the operating system can examine text content displayed within the application to predict text selection bounds along with a possible destination application for the selected text. Until the user initiates selection, there may be no indication that a piece of text content might be actionable. Further, the functionality may not work as intended in cases where application developers implement a custom operation for the input mode utilized for passing the text content and associated action from one application to another. With user permission, this disclosure applies regular expression parsing and neural network processing to the text portion of the on-screen content to detect text entities that might be actionable by the OS or other applications on the device. After merging the actionable text entities identified via either of the two techniques, the corresponding text is presented, e.g., by underlining the corresponding text and linking it to invoke the corresponding action.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.