Dianna Radpour

State University of New York at Buffalo


Hi there, and thanks for visiting my page. I am currently a graduate student at the University of Buffalo, pursuing a Masters in Computational Linguistics.

My research focuses primarily on exploring how neural networks can be implemented for natural language understanding, as well as in machine translation. Recently, I have been researching how deep learning methods can be used for sentiment analysis, specifically in sarcasm detection. I've also been working on creating parallel colloquial corpora to enhance the neural machine translation process between Farsi and English.

In my free time, I love baking, being in water, and graphic novels. I also enjoy collecting and filling my walls with my sister's photography. Check out her super spunky photos (like the one pictured here) at www.sepeadeh.com!


I recently gave a talk at the 2017 Sentiment Analysis Symposium in New York, titled "Sarcasm and its Symptoms." I spoke about my recent collaborative efforts with Vinay Ashokkumar, at the University at Buffalo, in developing an automatic sarcasm classifier. Our approach used reviews from the most recent Yelp dataset of 4.1 million reviews to introduce a novel probabilistic modeling framework for identifying, classifying and learning features of sarcastic text via training a neural network with human-informed sarcastic benchmarks.

This past January, I was invited to Amazon’s 2017 Graduate Research Symposium in Seattle for my research in designing and implementing a method to disambiguate between identical words being used in different contexts, specifically idiomatic versus literal. I created different corpora of each of the identical target word pairs as input for word2vec, for differentiating between word embeddings.



Feel free to contact me at diannara(at)buffalo.edu.