Just Act Natural: Berkeley students leverage language data models to help people sound more natural and check the formality of their writing
The information age has undoubtedly brought the world closer. People across the globe have become interconnected in a manner that was inconceivable even 30 years ago. With this it has brought the need for people to communicate in a common language and in a large part of the world that language has been English.
However, as non-native English speakers shift over to speaking and writing in English, while what they write may be grammatically correct, it may lack natural tone and flow.
While speaking with non-native English speakers seeking opportunities in America, it is clear that there is a strong desire to have their business communication be as formally written and natural sounding as
possible, in addition to being grammatically correct, so that they feel more secure that their current lack of proficiency in English doesn’t negatively impact their career lives. While this opportunity seems to have been missed by the traditional online grammar tool developers, a team of 6 UC Berkeley students have tackled the issue by developing a tool which allows users to construct emails with proper format, grammar, and tone.
The team of Berkeley students have developed and implemented cutting edge data driven language models that analyze everything from a passage’s tone, and formality to its basic grammar. The team started by collecting massive language datasets of the English language derived from dozens of online sources. Subsequently the team parsed, cleaned, and engineered features from that data to train advanced statistical and deep learning models to best fulfill the user’s needs. The final product is displayed on an interactive, easy to use platform where users can write their messages from scratch, receiving feedback as they go.
With this interactive platform users will be able to not only utilize these data driven insights to improve their writing, but the platform also offers templates and suggestions for how to structure everything from an academic email to a resignation letter. The platform will help users easily understand how to write naturally and will aid along every step of the process of crafting any text to be published online. In addition to this, the platform features an incredibly insightful formality checker that provides real time feedback that will allow users to calibrate the tone of their text. Ultimately, the Berkeley students hope that their tool will specifically help any non-native English speakers who may lack experience writing professional or formal emails thus enhancing their employability!
Project By: Aaron Lin, Hanson Qin, Chris Dong, Matthew Dodd, Tyler Riverbay, Steven Johannemann