Personal tools
User menu

NieDong

From Computational CyberPsychology Lab

Jump to: navigation, search

Dong Nie(聂栋)

Niedong.jpg

Profile

Research Interests

Machine Learning, Data Mining, Recommender Systems, Natural language processing

Education

  • M.S. Computer Science, University of Chinese Academy of Sciences, Beijing, China. 2011 - present.
  • B.S. Software Engineering, NorthEastern University, Shenyang, China. 2007.9 - 2011.7.
    • Qualified for postgraduate recommendation to THU, PKU and ZJU.

Research Experience

Undergraduate Research Projects

  • Learning data structure and basic algorithms about computer science and participating in ACM/ICPC contests.
  • Learning mathematics modeling and participating in modeling contests.
  • From May to August,2008,participating in National Information Security Competition
  • From fall semester, 2008 to next spring semester, being an intern in information integration Lab, studying work about web service
  • From fall semester, 2009 to next spring semester, learning things about linux and doing research in embedded system. Two papers were published in the period.
  • From June to August, 2010, being an intern at neusoft corporation, doing research on network communication in linux environment and learning something about image processing
  • Late August,2010, participating in National Students outsourcing innovation Application Contest

Graduate Research Projects (Feb 2011 - Present)

  • WebMind Browser (Feb, 2011-May, 2012)
    • Constructing a browsing system named WebMind, retrieving users' internet browsing behavior based on WebMind platform, extracting features, analyzing their internet behaviors and predicting people's psychology state in multiple dimensions. Then a psychology intervention recommender system is developed using content-based and collaborating filtering algorithm. In addtion, users' behavioral intention is also taken into consideration in our WebMind system. A page sequence pattern is used to identify users' intention and keywords are recongnized to represent users' intention. For this project, I implement a Chinese word segmentation program using double-array trie algorithm. I also implement a Naive Bayes Classifcation to deal with page content type. To identify the psychology traits, I turn to libSVM for help and I adjust the soft margin due to imbalance of the data.
  • Psychology Mining based On Social Media (June, 2012-Present)
    • This is the core research program of our laboratory. We mainly collect data from Weibo (China’s twitter) platform, and try to predict user psychology traits with the help of machine learning methods. I take part in the whole process of this research, including downloading data from social network, preprocessing data, feature extraction and modeling. My core work is to design and implement useful algorithms for my laboratory, such as semi-supervised learning and active learning algorithms. We have successfully developed mental health warning system, and we cooperate with university to develop a psychology site. In addition, we cooperate with Sina Weibo and publish a report about microblog users every week. We are also helping Yonyou (which is China’s SAP) construct a user experience intelligent evaluation system. I have published several papers during the research period.
  • Machine Learning Study (Sep.,2012-Present)
    • As our lab's data share the same feature with most scenes that labeled data is limited while unlabeled data is of large scale, I seek machine learning solutions to make use of unlabeled data. The methods I study are unsupservised learning, semi-supervised learning and active learning. I design and implement graph-based semi-supervised learning (SSL) methods for our data. I first implement a semi-supervised classification based on local and global consistency, later, I use graph Laplacian norm to enforce the smoothness constraint and implement a semi-supervised regression method. To challenge larger amounts of user study, we'd better choose the most useful subjects to take part in our experiments in advance. I turn to active learning regression methods, especially algorithms proposed by Masashi Sugiyama which aims to minimize the prediction error based on a pool of test input points. As the large scale of the data, cluster approaches are sometimes exploited to deal with data. Several feature selection methods have also been tested.
  • Recommender Systems (June, 2012-Present)
    • I take interest in recommender systems and conduct some simple research. I mainly focus on how to use unlabeled data to improve the performance of model-based recommender systems, at the same time, I am also interested in recommender systems based on social relationships. I have designed and implemented some recommender algorithms and have published one paper.
  • Sentiment Analysis (July, 2013- Sep., 2013)
    • This program is to observe the sentiment variation of microblogs before and after Yaan earthquake (2013-04-20). The Microblog statuses are from users who located in Yaan. A program similar to LIWC is used to extract features from the Microbog statuses, and about 900 statuses are rated from -10 to 10 for sentiment the statuses express, manifold-ranking algorithm is used to predict the sentiment score of rest statuses (we also use SVR). Then we use a greedy algorithm to calculate the score for a user. Eventually, the sentiment variation comes out in this way.

Publications

  • Dong Nie, Shuotian Bai, Bibo Hao and Tingshao Zhu. Predicting Personality on Social Media with Active Learning. WSDM 2014. (In Review).
  • Dong Nie, Lin Li and Tingshao Zhu. Predicting Personality on Social Media with Semi-Supervised Regression. SDM 2014. (In Review).
  • Dong Nie, LinZi Hong and Tingshao Zhu. Movie Recommendation with Genre Preference Regularization. ICMLA 2013, Miami, America.
  • Dong Nie, Zengda Guan, Ang Li and Tingshao Zhu. Your Search and Your Personality. ICPCA 2013, Chile.
  • Dong Nie, Lin Li and Tingshao Zhu. Conscientiousness Measurement from Weibo’s Public Information. Partial Supervised Learning 2013, NanJing.
  • Dong Nie, Ang Li, Bibo Hao and Tingshao Zhu. Personality Traits and Microblogging Behavior of Weibo Users: Onlies versus Others. ACM Web Science 2013. May 2013, Paris.
  • Dong Nie, Yue Ning and Tingshao Zhu. Predicting Mental Health Status in the Context of Web Browsing. WI workshop on Web Personalization, Recommender Systems and Social Media, 2012, Macau.
  • Xinshuai Che, Xu Bai, Dong Nie and Chunyan Han. An Automobile Emergency Calling System, ICCDA, 2010.
  • Xinshuai Che, Dong Nie and Chunyan Han. Alarm system for Greenhouse temperature management, ICSPS, 2010.

Professional Activities

  • Speaker and Student Volunteer in Web Intelligence Congress, 2012, Macao.
  • Speaker in MCS-PSL 2013, Nanjing, China

Computing Skills

  • I like C++ and matlab most, while I also use Java, and C, I am now learning python.
  • The IDE I often use is Visual Studio 2010 and Eclipse 4.3.
  • I master data structure and algorithms well.

Honors and Certificate

  • First-class (1 times), second-class (5 times) and third-class (1 time) learning scholarship (university-level) during 2007-2011.
  • First Prize in Northeastern University mathematical contest of modeling, 2009 and third prize in University MCM (Mathematical Contest in Modeling, America), 2010.
  • Second prize in University On-Site algorithm contest, 2010. Second prize in ACM/ICPC Liaoning Provincial On-Site Contest, 2010 and Third prize in Multi-provincial On-Site Contest, 2010.
  • Third prize in National Students outsourcing innovation Application Contest.
  • Linux CT Certificate.


Hobbies

PE, Culture and Writing.

  • This page was last modified on 30 October 2013, at 16:11.
  • This page has been accessed 25,797 times.