Tingting Feng
Research Subjects
Tingting Feng has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Yuchun Guo
Beijing Jiaotong University
2 shared publications Baijun Shen
PPLive Inc. Shanghai, China
1 shared publicationShan Yang
China Unicom Smart City Research Institute
1 shared publicationTi Wang
China Unicom Smart City Research Institute
1 shared publication- Ting Xu
Center for the Integrative Developmental Neuroscience, Child Mind Institute, New York, USA
1 shared publication - Wei Zhu
School of Materials Science and Engineering, Beihang University, Beijing, China
1 shared publication
Latest Publications
- 26 January 2020
- Published by Springer Nature
- p. 345-356
Any Privacy Risk if Nobody’s Personal Information Being Collected?
- 26 January 2020
- Published by Springer Nature
- p. 345-356
Any Privacy Risk if Nobody’s Personal Information Being Collected?
- 1 September 2015
- Published by Elsevier in Computer Communications
- Vol. 68, 17-24
Can user privacy and recommendation performance be preserved simultaneously?
- 1 September 2015
- Published by Elsevier in Computer Communications
- Vol. 68, 17-24
Can user privacy and recommendation performance be preserved simultaneously?
- 1 June 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 12 (15503607) , 1837-1842
Tags and titles of videos you watched tell your gender
In online video systems, viewer demographic information (gender, age, etc.) is of huge commercial value for delivering targeted advertising and video recommendations, but generally not available directly. This paper targets inferring viewers' gender based on implicit watching history in the large-scale online video systems. To tackle the sparsity problem without filtering out any cold users or videos, we not only introduce video tags as features, but also use an efficient Chinese word segmentation method to extract hot key-words from video titles as features. Moreover, users' viewing behavior distribute lognormally, hence we apply a logarithmic transformation on the inference matrixes and further find key features via principal components analysis (PCA). We then solve the gender inference as a classification problem and define some modified evaluation metrics adapt to the imbalance classification problem. We compare a set of classifiers including Class prior, EM, SVM, Logistic regression, Partially supervised soft-label and belief-based mixture and find that Logistic regression is the best. The inference results show that our algorithms can obtain high F̃ 1 values for all classes. The highest value of PPTV dataset can reach nearly 0.75. And inference based on key-words results in a 14.63% increase of F̃ 1 contrast to the ratings of MovieLens.
- 1 June 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 12 (15503607) , 1837-1842
Tags and titles of videos you watched tell your gender
In online video systems, viewer demographic information (gender, age, etc.) is of huge commercial value for delivering targeted advertising and video recommendations, but generally not available directly. This paper targets inferring viewers' gender based on implicit watching history in the large-scale online video systems. To tackle the sparsity problem without filtering out any cold users or videos, we not only introduce video tags as features, but also use an efficient Chinese word segmentation method to extract hot key-words from video titles as features. Moreover, users' viewing behavior distribute lognormally, hence we apply a logarithmic transformation on the inference matrixes and further find key features via principal components analysis (PCA). We then solve the gender inference as a classification problem and define some modified evaluation metrics adapt to the imbalance classification problem. We compare a set of classifiers including Class prior, EM, SVM, Logistic regression, Partially supervised soft-label and belief-based mixture and find that Logistic regression is the best. The inference results show that our algorithms can obtain high F̃ 1 values for all classes. The highest value of PPTV dataset can reach nearly 0.75. And inference based on key-words results in a 14.63% increase of F̃ 1 contrast to the ratings of MovieLens.