UROP Project

Automatic Assessment of Text Complexity

Advances in natural language processing have paved the way for the development of computational tools designed to automatically assess the complexity of spoken and written language samples. There are a variety of computational tools available that afford speed, flexibility and reliability and permit the direct comparison of numerous indices of language complexity. Considerable gains have been made from the use of such tools across a variety of application areas, such as (1) automated assessment of second language (L2) writing performance and proficiency, (2) automated assessment of L2 spoken performance and proficiency, (3) detection and prevention of fake news, (4) stance detection in social media, etc. However, all these available tools only generate for each complex measure a single score (summary statistics) that represents the global complexity of a language sample. We have developed a computational tool that implements a sliding-window approach to track changes in complexity within a language sample derived from a series of measurements obtained by the tool.

Task

In its current version, the tool supports 154 complexity measures derived from language acquisition and processing research. The tool was designed with extensibility in mind, so that additional complexity measures can easily be integrated. Your task is to work on the further development of the tool including the integration of new complexity measures and the development of a platform independent desktop application. You will also perform text classification experiments in one of the application areas mentioned above.

Requirements

Interest in natural language processing, good programming skills ideally in JavaScript or Python