HUNG CHAU's blog

Professor Louis-Philippe Morency from CMU presented the talk "Modeling Human Communication Dynamics" on October, 21, 2016. at CS department, University of Pittsburgh. He started the talk with an introduction about his research that focuses on creating computational technologies to analyze, recognize and predict human subtle communicative behaviors in social context.  Speaking of human communicative behaviors, he indicated three main aspects, namely, verbal, vocal and visual aspects. There are four challenges in modeling human communication dynamics as stated by prof. Morency, which are behavioral, multimodal, interpersonal, and societal dynamics. He suggested that the model can broadly apply to healthcare, education, and online multimedia. In the three main parts of the talk, he performed his group's recent achievements, mutlimodal machine learning, and predicting listener behaviors. Firstly, he gave a demo about a healthcare decision support system using Multisense

Cristian Danescu-Niculescu-Mizil from University of Cornell gave the talk "Language and Social Dynamics" on September, 30, 2016 at CMU. He presented his past work including exploring the relation between users and their community, a computational framework for identifying and characterizing politeness, and predicting the future evolution of a dyadic relationship. At the beginning, Cristian talked about the linguistic change of users when they get involved in a community. He showed that users follow a determined life-cycle regarding to the process they adopt new community norms. When taking part in a community, new members adapt to existing community norms. For a long time, members also may adapt new norms or be innovators themselves, setting new trends. Others may keep their old styles and have no reaction to the change. Those members are more likely to leave their communities. Based on this assumption, the system can predict how long they will stay active in the community.

Dipanjan Das from Google presented the talk "Two Case Studies in Semantic Inference" on October, 14, 2016. He performed the two different semantic inference tasks. The first one focuses on the structure of natural language questions. And, the other is about more unstructured forms. Firstly, Dipanjan described a method for parsing natural language questions to logical forms. These logical forms can be softly mapped to the information stored in the structured knowledge base. And then the system matches the forms (sub-graphs) from knowledge base to Question Answer mechanism. In the semantic parsing process, he presented a Dependency Parser that extract sentences' structure, and DepLambda to parse into logical forms, which is based on lambda calculus.  Dipanjan carried out an empirical study to compares DepLambda techniques to other baselines. He showed that DepLambda, in two test collections, performs better than Simple Graph, CCG Graph and Deptree. In t

Lingjia Deng from Intelligent Systems Program, University of Pittsburgh had a talk about Entity/Event-level Sentiment Detection and Inference on October, 07, 2016. Lingjia's work focused on sentiment analysis and opinion mining. She introduced an sentiment analysis model that aims at detecting both explicit and implicit sentiments expressed among entities and events in text. As stated in the talk, most of the work in opinion mining focuses on extracting explicit sentiments. For example, in the sentence "she loves traveling", the sentiment that is explicitly expressed is positive. However, in another example, "People celebrate that Trump was defeated." there are two opinions in this sentence. One is also positive, based on the word celebrate , people celebrate something . The other is negative, based on the word defeated , Trump was defeated . But besides these explicit opinions, we can infer implicitly another sentiment in the sentence which is that people i

I found this article on the Facebook of a professor. After reading through the whole content, I have learned some useful experience from the author. I quote some ideas here: "There are very few people who make it to the top PhD programs. You’d be joining a group of a few hundred distinguished individuals in contrast to a few tens of thousands (?) that will join some company." (just personal opinion. Don't condemn :) ) "As a PhD student you’re your own boss. Want to sleep in today? Sure. Want to skip a day and go on a vacation? Sure. All that matters is your final output and no one will force you to clock in from 9am to 5pm. Of course, some advisers might be more or less flexible about it and some companies might be as well, but it’s a true first order statement." (just motivate myself) "You will inevitably find yourself working very hard (especially before paper deadlines). You need to be okay with the suffering and have enough mental stamina and dete