Machine Learning for NLP: Seminars (2015)

Instructions

The presentations are done individually or in pairs. If you choose to work individually, it's OK if the presentation is around 15 minutes; if you work in pairs, it should be a bit longer.

Read through the paper(s) you have selected, and try to think how to present the main ideas to your classmates. Here are some useful points that you can keep in mind when you make your presentation. Obviously it's useful if you go through the main parts of the paper:

• What is the task that they are trying to solve? (That is, what kind of system are they trying to build?)
• What are the questions that they are investigating?
• How do they solve the problem? How is the system implemented?
• What kind of experiments do they carry out? What data did they use? How did the evaluation work?
• What results did they get, and what conclusions do they draw?

In addition, it could interesting if you could also try a bit to read the paper with a critical eye. Here are some things that you could think of:

• If you were given the task of implementing a system like this, how difficult do you think it would be? How much time and what kind of resources would you need?
• Are there any things that you think you could do better than the authors do?
• Is there anything in the paper (implementation, experiments, conclusion) that you disagree with?
• What do you think of the linguistic assumptions that they are using in their system?
• Any other interesting comment?

Ideally, you prepare slides electronically (preferably a pdf), but it's also OK to use the whiteboard.

List of seminars

• (Sep 11) Florian and Bianka discussed Thumbs up? Sentiment classification using machine learning techniques by Pang, Lee, and Vaithyanathan (2002).
• (Sep 18) Karin and Tessa discussed The ups and downs of preposition error detection in ESL writing by Tetreault and Chodorow (2008).
• (Sep 18) Liina and Luke discussed Behavioral-based cheating detection in online first person shooters using machine learning techniques by Alayed, Frangoudes, and Neuman (2013).
• (Sep 25) Bob discussed A readable read: Automatic assessment of language learning materials based on linguistic complexity by Pilán, Vajjala, and Volodina (2015).
• (Sep 25) Wafia discussed Sentiment analysis for Modern Standard Arabic and colloquial by Ibrahim, Abdou, and Gheith (2015).
• (Oct 2) Sophie discussed Multi-document summarization by maximizing informative content-words by Yih, Goodman, Vanderwende, and Suzuki (2007).
• (Oct 9) Isac discussed Reinforcement learning for spoken dialogues systems by Kearns, Litman, and Walker (1999).
• (Oct 9) Anna discussed Automatic labeling of semantic roles by Gildea and Jurafsky (2002).
• (Oct 16) Haixia discussed Chinese word segmentation as character tagging by Xue (2003).
• (Oct 16) Yuri discussed Distributed representations of words and phrases and their compositionality by Mikolov, Sutskever, Chen, Corrado, and Dean (2013).
• (Oct 19) Toni will present Word representations: A simple and general method for semi-supervised learning by Turian, Ratinov, and Bengio (2010).