Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userPrivacy Guarantees for De-identifying Text Transformations
124
0
0.0
(
0
)
Added by
David Adelani
Publication date
2020
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Machine Learning approaches to Natural Language Processing tasks benefit from a comprehensive collection of real-life user data. At the same time, there is a clear need for protecting the privacy of the users whose data is collected and processed. For text collections, such as, e.g., transcripts of voice interactions or patient records, replacing sensitive parts with benign alternatives can provide de-identification. However, how much privacy is actually guaranteed by such text transformations, and are the resulting texts still useful for machine learning? In this paper, we derive formal privacy guarantees for general text transformation-based de-identification methods on the basis of Differential Privacy. We also measure the effect that different ways of masking private information in dialog transcripts have on a subsequent machine learning task. To this end, we formulate different masking strategies and compare their privacy-utility trade-offs. In particular, we compare a simple redact approach with more sophisticated word-by-word replacement using deep learning models on multiple natural language understanding tasks like named entity recognition, intent detection, and dialog act classification. We find that only word-by-word replacement is robust against performance drops in various tasks.
rate research
Read More
Texts convey sophisticated knowledge. However, texts also convey sensitive information. Despite the success of general-purpose language models and domain-specific mechanisms with differential privacy (DP), existing text sanitization mechanisms still provide low utility, as cursed by the high-dimensional text representation. The companion issue of utilizing sanitized texts for downstream analytics is also under-explored. This paper takes a direct approach to text sanitization. Our insight is to consider both sensitivity and similarity via our new local DP notion. The sanitized texts also contribute to our sanitization-aware pretraining and fine-tuning, enabling privacy-preserving natural language processing over the BERT language model with promising utility. Surprisingly, the high utility does not boost up the success rate of inference attacks.
Sentiment analysis is a text mining task that determines the polarity of a given text, i.e., its positiveness or negativeness. Recently, it has received a lot of attention given the interest in opinion mining in micro-blogging platforms. These new forms of textual expressions present new challenges to analyze text given the use of slang, orthographic and grammatical errors, among others. Along with these challenges, a practical sentiment classifier should be able to handle efficiently large workloads. The aim of this research is to identify which text transformations (lemmatization, stemming, entity removal, among others), tokenizers (e.g., words $n$-grams), and tokens weighting schemes impact the most the accuracy of a classifier (Support Vector Machine) trained on two Spanish corpus. The methodology used is to exhaustively analyze all the combinations of the text transformations and their respective parameters to find out which characteristics the best performing classifiers have in common. Furthermore, among the different text transformations studied, we introduce a novel approach based on the combination of word based $n$-grams and character based $q$-grams. The results show that this novel combination of words and characters produces a classifier that outperforms the traditional word based combination by $11.17%$ and $5.62%$ on the INEGI and TASS15 dataset, respectively.
Written text often provides sufficient clues to identify the author, their gender, age, and other important attributes. Consequently, the authorship of training and evaluation corpora can have unforeseen impacts, including differing model performance for different user groups, as well as privacy implications. In this paper, we propose an approach to explicitly obscure important author characteristics at training time, such that representations learned are invariant to these attributes. Evaluating on two tasks, we show that this leads to increased privacy in the learned representations, as well as more robust models to varying evaluation conditions, including out-of-domain corpora.
Recognizing that even correct translations are not always semantically equivalent, we automatically detect meaning divergences in parallel sentence pairs with a deep neural model of bilingual semantic similarity which can be trained for any parallel corpus without any manual annotation. We show that our semantic model detects divergences more accurately than models based on surface features derived from word alignments, and that these divergences matter for neural machine translation.
The correlations and network structure amongst individuals in datasets today---whether explicitly articulated, or deduced from biological or behavioral connections---pose new issues around privacy guarantees, because of inferences that can be made about one individual from anothers data. This motivates quantifying privacy in networked contexts in terms of inferential privacy---which measures the change in beliefs about an individuals data from the result of a computation---as originally proposed by Dalenius in the 1970s. Inferential privacy is implied by differential privacy when data are independent, but can be much worse when data are correlated; indeed, simple examples, as well as a general impossibility theorem of Dwork and Naor, preclude the possibility of achieving non-trivial inferential privacy when the adversary can have arbitrary auxiliary information. In this paper, we ask how differential privacy guarantees translate to guarantees on inferential privacy in networked contexts: specifically, under what limitations on the adversarys information about correlations, modeled as a prior distribution over datasets, can we deduce an inferential guarantee from a differential one? We prove two main results. The first result pertains to distributions that satisfy a natural positive-affiliation condition, and gives an upper bound on the inferential privacy guarantee for any differentially private mechanism. This upper bound is matched by a simple mechanism that adds Laplace noise to the sum of the data. The second result pertains to distributions that have weak correlations, defined in terms of a suitable influence matrix. The result provides an upper bound for inferential privacy in terms of the differential privacy parameter and the spectral norm of this matrix.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
