arXiv subjects

Benjamin C. M. Fung

Publications and source records attributed to Benjamin C. M. Fung.

21 records · Page 2Linked to original sources

arXiv2016

Learning Stylometric Representations for Authorship Analysis

Authorship analysis (AA) is the study of unveiling the hidden properties of authors from a body of exponentially exploding textual data. It extracts an author's identity and sociolinguistic characteristics based on the reflected writing styles in the text. It is an essential process for various areas, such as cybercrime investigation, psycholinguistics, political socialization, etc. However, most of the previous techniques critically depend on the manual feature engineering process. Consequently, the choice of feature set has been shown to be scenario- or dataset-dependent. In this paper, to mimic the human sentence composition process using a neural network approach, we propose to incorporate different categories of linguistic features into distributed representation of words in order to learn simultaneously the writing style representations based on unlabeled texts for authorship analysis. In particular, the proposed models allow topical, lexical, syntactical, and character-level feature vectors of each document to be extracted as stylometrics. We evaluate the performance of our approach on the problems of authorship characterization and authorship verification with the Twitter, novel, and essay datasets. The experiments suggest that our proposed text representation outperforms the bag-of-lexical-n-grams, Latent Dirichlet Allocation, Latent Semantic Analysis, PVDM, PVDBOW, and word2vec representations.

cs.CL↗

arXiv2012

Privacy Preserving Record Linkage via grams Projections

Record linkage has been extensively used in various data mining applications involving sharing data. While the amount of available data is growing, the concern of disclosing sensitive information poses the problem of utility vs privacy. In this paper, we study the problem of private record linkage via secure data transformations. In contrast to the existing techniques in this area, we propose a novel approach that provides strong privacy guarantees under the formal framework of differential privacy. We develop an embedding strategy based on frequent variable length grams mined in a private way from the original data. We also introduce personalized threshold for matching individual records in the embedded space which achieves better linkage accuracy than the existing global threshold approach. Compared with the state-of-the-art secure matching schema, our approach provides formal, provable privacy guarantees and achieves better scalability while providing comparable utility.

cs.DB↗

arXiv2011

Differentially Private Trajectory Data Publication

With the increasing prevalence of location-aware devices, trajectory data has been generated and collected in various application domains. Trajectory data carries rich information that is useful for many data analysis tasks. Yet, improper publishing and use of trajectory data could jeopardize individual privacy. However, it has been shown that existing privacy-preserving trajectory data publishing methods derived from partition-based privacy models, for example k-anonymity, are unable to provide sufficient privacy protection. In this paper, motivated by the data publishing scenario at the Societe de transport de Montreal (STM), the public transit agency in Montreal area, we study the problem of publishing trajectory data under the rigorous differential privacy model. We propose an efficient data-dependent yet differentially private sanitization algorithm, which is applicable to different types of trajectory data. The efficiency of our approach comes from adaptively narrowing down the output domain by building a noisy prefix tree based on the underlying data. Moreover, as a post-processing step, we make use of the inherent constraints of a prefix tree to conduct constrained inferences, which lead to better utility. This is the first paper to introduce a practical solution for publishing large volume of trajectory data under differential privacy. We examine the utility of sanitized data in terms of count queries and frequent sequential pattern mining. Extensive experiments on real-life trajectory data from the STM demonstrate that our approach maintains high utility and is scalable to large trajectory datasets.

cs.DB↗