This is the documentation for the Privacy module of SOFIE's PDS component.
This module enables the creation of privacy preserving surveys. These are surveys that allow users to add noise to their responses, using local differential privacy mechanisms. The addition of the noise prevents 3rd parties from learning meaningful information about specific users, but at the same time aggregated statistics can be extracted. The accuracy of the extracted statistics depends on the number of responses.
User results are stored in a smart contract, included in the module. Therefore, since all answers are public anybody can extract the aggregated statistics.
The architecture of the privacy module is illustrated in Figure 1. It is composed of a Service Provider that creates surveys, a Data Provider that provides responses to the survey, and a Statistics Consumer that extracts statistics using the provided responses. All responses are stored in an Ethereum smart contract.
Figure 1: The privacy module of the PDS component.
The privacy module is part of SOFIE's PDS component. PDS is used by SOFIE's Energy data exchange and Food supply chain pilots.
More information about this component and its relation to the SOFIE project can be found in D2.5 Federation Framework, SOFIE deliverable
The software modules are implemented in Python 3. Survey responses are stored in an Ethereum smart contract. In order to provide local differential the basic one-time RAPPOR algorithm is used. This algorithm is described in
U. Erlingsson, V. Pihur, and A. Korolova, “RAPPOR: Randomized Aggregatable Privacy-Preserving Ordinal Response,” in Proc. of ACM SIGSAC Conference on Computer and Communications Security, 2014
Python 3 and the web3 library are required. Web3 can be installed Ubuntu 18.04 using
pip3 install web3Compile and deploy the contract\survey.sol smart contract.
Edit the configuration file conf/privacy.conf and provide the following information.
| Parameter | Meaning |
|---|---|
| web3provider | The URL of the Web3 provider |
| survey_sc_address | The address of the deployed survey contract |
Before using this module a survey must be created. Each survey is composed of a single question that has multiple possible answers (multiple choice question).
A survey can be used using the Service_Provider class included in the service_provider.py file.
This class implements a method called create_survey that accepts the following inputs:
| Parameter | Meaning |
|---|---|
| number_of_choices | (int) The number of possible choices to the question |
| survey_name | (string) the name of the survey |
A user can privately respond to a survey by using the Provider class included in the data_provider.py file. This class implements
a method called record_response. This method accepts the following inputs:
| Parameter | Meaning |
|---|---|
| number_of_choices | (int) The number of possible choices to the question |
| correct_choice | (int in the range of [0, number_of_choices)) the correct choice |
| survey_name | (string) the name of the survey |
This method generates a privacy preserving response and stores it in the smart contract.
If a significant number of responses has been collected(~1000) then statistics with quite good precision can be collected. The class Consumer
implemented in the statistics_consumer.py file, provides a method (estimate_responses) that generates the probability of each choice, based on the information
recorded in the smart contact. This method accepts the following inputs:
| Parameter | Meaning |
|---|---|
| number_of_choices | (int) The number of possible choices to the question |
| survey_name | (string) the name of the survey |
It returns an array of size number_of_choices and each element of the array is the probability of the corresponding choice.
For testing purposes ganache-cli is required. To install ganache execute:
npm install -g ganache-cliTests are executed using pytest and pytest-asyncio. To install it execute
pip3 install -U pytest
pip3 install pytest-asyncioExecute the following command
python3 -m pytest -s tests/privacy/test_privacy.py Test results are shown in the standard output.
No known issues
Please use github issues for posting questions.
This component is licensed under the Apache License 2.0.