Today was the last day of the main conference of the Annual Meeting of Association for Computational Linguistics (ACL). It is my first ACL. Thanks to my professor Dr. Thamar Solorio for supporting me to attend this amazing event. I’ve attended some amazing talks, had a chance to visit some posters with cool ideas, talked to some great people. Just came back to hotel and thought lets write about some memories?
Because of my primary interests in semantics, sentiments and information extraction I’ve attended these sessions. Here are the talks I found so exciting and marked to read the papers soon and possibly use in my work.
- Skip-Gram – Zipf + Uniform = Vector Additivity
- The State of the Art in Semantic Representation. (Actually I read this paper before I came to ACL. Its a really nice survey paper discussing existing systems for semantic representation of texts)
- Learning Cognitive Features from Gaze Data for Sentiment and Sarcasm Classification using Convolutional Neural Network.
- Overcoming Language Variation in Sentiment Analysis (They used social network connection to analyze sentiments. Isn’t it a cool thing to do?)
- Abstract Meaning Representation Parsing using LSTM Recurrent Neural Networks. (Before coming to ACL I was working on CAMR and this new system did improvement over it. Is there any reason not to like it? 😀 Althought the source code isn’t released yet, but the author said he will release it soon. Waiting to use it.)
- Neural Architectures for Multilingual Semantic Parsing. (Have to try it too)
- Deep Pyramid Convolutional Neural Networks for Text Categorization. (I know a lot of new architectures are drafted in some latex documents to be published soon. But still, new things are fun to explore.)
- Beyond Binary Labels: Political Ideology Prediction of Twitter Users.
- Leveraging Behavioral and Social Information for Weakly Supervised Collective Classification of Political Discourse on Twitter.
– I will remember these two talks for many days. Such practical scenarios. Presenter of the second talk said during the question answering part ‘Democrats are kind of scattered about their choice of topics. But the republicans maintain a strong and stable point of topics.’ – We know its true.
- Context-Dependent Sentiment Analysis in User-Generated Videos. (Deep learning everywhere :/ )
- Apples to Apples: Learning Semantics of Common Entities Through a Novel Comprehension Task.
- Enriching Word Vectors with Subword Information. (Check it guys. Fasttext is the name of the system.)
- A Transition-Based Directed Acyclic Graph Parser for UCCA.
- A Corpus of Compositional Language for Visual Reasoning. (I talked with the first author Alane Suhr at the aqurium yesterday. So nice person with nice works. She won the best paper. Congratulations to her 🙂 )
In the poster session, there were several interesting works. I liked the poster of ‘Parser Adaptation for Social Media by Integrating Normalization.’ There were a lot of works I found so cool but its difficult to summarize the long list now. So tired basically. I enjoyed both of the keynotes, the opening speech. A lot of things to learn.
In these three days I met with some people the NLP community really love for their works. Christopher Manning, Dan Jurafsky, Raymon Mooney, Noah Smith,, Kevin Knight, Martha Palmer, Yoav Goldberg, Alan Ritter, Saif M. Mohammad.
In recent days I followed works from Yoav (His blog in Medium was mind blowing. Thanks to Dr. Solorio; she gave the blog to read) and Saif and I had in mind that I had to say at least Hi to them. But they were so nice that eventually I took pictures with them. Saif gave some great ideas for my current work. Thank you so much.
So thats all for now. Vancouver is really a nice city with great people. I met a bus driver (Charles Dixon) here. He is the coolest driver I’ve met (All fans of The Beatles are cool; no question about it). I was surprised to know he was severely assaulted by a guy in 2011. This world is mysterious, so as the humans (sigh). Here I am with him in his bus.
So far it was a great experience and I got my fuel to work after I go back to Houston. I am already missing my lab.
I will be presenting a poster (Sentiment Analysis on Financial Data Using Neural Network) in SemEval (Task 5) this saturday. Hope it will be another day full of motivations and learning experience from the community.
Finally, I didn’t know that my lab has a restaurant branch in Vancouver.