1 . (10%) State the problem the paper is intending to solve. This kind of paper is intending to demonstrate how Airavat, a MapReduce-based system for allocated computations supplies end-to-end confidentiality, integrity, and privacy warranties using a combination of mandatory get control and differential personal privacy which provides reliability and privacy guarantees against data leakage. 2 . (20%) State the main contribution with the paper: solving a new problem, proposing a fresh algorithm, or perhaps presenting a fresh evaluation (analysis).
If a new problem, why was the issue important? Is the problem continue to important today? Will the issue be important another day? If a fresh algorithm or perhaps new evaluation (analysis), exactly what are the advancements over past algorithms or evaluations?
Just how do they come program the new algorithm or evaluation? The main contribution of the newspaper is that Airavat builds on mandatory access control (MAC) and differential privacy to make certain untrusted MapReduce computations upon sensitive data do not leak private information and offer confidentiality, honesty, and level of privacy guarantees. The goal is to prevent malicious computation providers from breaking privacy guidelines a data supplier imposes around the data to avoid leaking details about individual items in the info.
The system is usually implemented as a modification to MapReduce as well as the Java virtual machine, and runs on top of SELinux a few. (15%) Sum it up the (at most) several key key ideas (each in 1 sentence. ) (1) 1st work to include MAC and differential privateness to mapreduce. (2) Offers a new structure for privateness preserving mapreduce computations. (3) Confines untrusted code. This technique provides reliability and privateness guarantees intended for distributed computations on sensitive data with the ends. However , the data continue to can be leaked in the impair.
Because multiple machines are involved in the calculation and malicious worker can sent the intermediate data to the outdoors system, which will threatens the privacy of the input data. Even to not this degree, temporary info is kept in the workers and people data may be fetched also after computation is done. m. Rate how convincing the methodology is definitely: how do the authors warrant the solution approach or analysis? Do the authors use disputes, analyses, tests, simulations, or a combination of all of them?
Do the claims and a conclusion follow in the arguments, analyses or tests? Are the assumptions realistic (at the time from the research)? Will be the assumptions nonetheless valid today? Are the tests well designed? Is there different tests that would be even more convincing?
Are available other alternatives the experts should have regarded as? (And, of course , is the conventional paper free of methodological errors. ) As the author’s stated on page three or more “We try to prevent malicious computation providers from violating the privacy policy of the info provider(s) by leaking info on individual data items. ” They use gear privacy device to ensure this kind of. One interesting solution to data leakage is they have the mapper specify a range of its keys. It appears like that the larger your data set is, the more privacy you may have because a end user affects fewer of the end result, if removed.
They revealed results that have been really close to 100% together with the added noises, it seems this can be viable strategy to protect the privacy of the data input c. What is the most important limitation of the procedure? As the authors talk about, one computation provider could exhaust this budget over a dataset for a lot of other calculation providers and use a lot more than its fair share. While there is definitely some evaluation of successful parameters, a few large number of variables that must be set for Airavat to function properly.
This kind of increases the likelihood of misconfigurations or configuration settings that might seriously limit the computations which can be performed around the data. a few. (15%) What lessons ought to researchers and builders take away from this work. What (if any) inquiries does this work leave available? The current rendering of Airavat supports both equally trusted and untrusted Mappers, but Reducers must be trusted and they also customized the JVM to make mappers independent (using invocation figures to identify current and previous mappers).
They also modified the reducer to provide differential privacy. From the data provider’s perspective they have to provide many privacy guidelines like- privateness group and privacy finances. 6. (10%) Propose the improvement on a single problem.
I possess no recommended improvements.
