This summer I worked with Prof. Brian King, Computer Science, to develop an improved word prediction algorithm for users of speech generating devices. These devices fall into a field that is known as Augmentative and Alternative Communication (AAC) - a field that comprises any communication that is non-verbal, including hand gestures, facial expressions and writing. The information that we gathered about speech generating devices suggested one significant downfall - the devices severely reduce the speed at which one of their users is able to communicate. We saw this as an opportunity to develop a mechanism that would help increase the communication rate of users of speech generating devices. The mechanism that we decided to pursue was to create a word prediction algorithm, like what you see when you use Google, in order to decrease the time a user spends typing the messages they want to convey, and thereby increasing the speed at which they are able to communicate.
The best part about this research was its collaborative nature. Our project was particularly engaging because it required that we step out of our comfort zones; neither of us is explicitly knowledgeable about language modeling; however our shared interest in computer science provided us a base from which we were able to build. This base allowed us to combine techniques from our separate areas of interests, Prof. King's expertise in bioinformatics and my pursuits as a linguistics major, to solve the problem.
Related Reading: Rewards of Research