In Summer 2017, I worked with Professor Dragomir Radev in the Language, Information, and Learning at Yale (LILY) Lab to develop the All About NLP (AAN) interface. The site evolved from the ACL Anthology Network (conveniently also AAN) which parses citation information from papers in the ACL Anthology and analyzes their citation network. This summer, our goal was to create a new section of the site for hand-curated natural language processing (NLP) teaching resources.
Before starting, the first step was to look at the existing code. Since the website has been maintained by different people each year, there did not seem to be a method to the madness with all the PHP files scattered in the same directory. Thus, my first step was to comb through the existing code and organize everything into an MVC framework. In addition to optimizing the code, this also makes it possible for the code to remain organized in the future.
After that, I implemented the new features, some of which are shown below:
Now that the current platform is in place, we can incorporate more AI components to the system. As foreshadowed in the third screenshot, I am currently working on a research project to automatically generate coherent surveys for a given topic based on the contents of available resources in the system. This is essentially like multi-document summarization except with multiple resource types from research papers to tutorials on the internet.