Data Privacy and Protection

With the exponential growth   of use of big data analytics the privacy and ethical issues related to the use of the datasets and insights from the data are also raising concerns. One of the best online streaming services released their viewer dataset as the part of the data science challenge in order to improve its recommendation engine (Papernot & Goodfellow, 2018).   This challenges raised privacy and ethical concerns as according to some researchers it contains micro-data which is helpful in determining or specifying a user/individual.  

This kind of data about specific individuals, are gradually becoming available in the public domain due to the “open government” legislation or data mining research field. This kind of dataset often includes individual preferences and transactions (such as movie genre preferences in case of Netflix) that some people may consider as private/ sensitive. The following report contributes to the analysis and discussion about the different privacy and ethical issues related to the data analytics on the datasets that are publically available.  

Privacy of data is concerned with the protection of it as well as access to any individual’s information. This definition also includes the independence from the surveillance or unwanted attention by any government or business organization (Sankar & Parker, 2017). Privacy also implies that individuals or the users are the governor of their personal data and must be informed about the use, update and deletion of their data. Even after that only few people have actual control over their data which includes what data is stored about them, its accuracy and use of their data for different intents. 

Privacy of data in analytics is dependent on the organization that is collecting the data. It is intended that that the data must be used solely for the purpose it is collected stated to user and nothing else from that particular.

It is also desired that the data should not be shared between different business for generating profits or penetrate in a market. On the other hand, there are too many tools, techniques are used in the big data analytics that are helpful in the extraction of the user private data that violates the privacy rights of the user’s (Papernot & Goodfellow, 2018).  In addition to that, the security policies need to be enforced with privacy policy in order to protect all the entities involved in the process.

From the different researches in the field of the dataset it is evident that, with the datasets that does not include the personal identifiers such as name, unique customer or identification number it is possible to find out the preferences if the attacker had knowledge about a specific individual and their likes and dislikes for different genre of movies.

Risk of Identification and Disclosure

According to the authors Vayena et al.(2015), in case any analyst  previously knows someone and their likes and dislikes for the movies then using the available Netflix data set the researcher/attacker gain knowledge about their viewing history until the year 2005.

The researchers also answered some of the questions that encounters the possibility of   identification of the user only through available   dataset provided by Netflix.  In the response of this this is argued that there are other data sources that may help in matching the user preferences on multiple platforms such as IMdb and MovieLens. Even though the users on this platform may not use their name and actual identity on these sites but with the help of cross co-relation mechanisms it is possible to find out identity of the user (Sankar & Parker, 2017).  The complete process depicts that any attacker can find out links with any anonymous Netflix data record to any external publically available data of an individual such as their IMDb ratings that are associated with identity of a specific person. While Netflix’s exposed data sources excludes user /subscriber names, and used anonymous identifier for their subscribers/users but it is found by the   researcher’s that collection of movie ratings collective with available public database of ratings can easily detect or help in identifying people. Hajian, Bonchi and Castillo (2016), described the privacy issues that can be raised by using public reviews published by a different common user (who uses both Netflix and IMDb) on the Internet Movie Database (IMDb).

Exposing movie ratings posted by the user’s /reviewer wo though those data will be private could expose significant details about the person when analysed with other platform. For example, the researchers found that some people had strong and private opinions for the liberal and gay-themed films that are loved by some other groups.

In different other sectors, there are example of use of the analytics and breaches that harms the privacy the owner/ individuals about whom the data was collected.  The impact also includes different national privacy regulations and policies.  Even though the use of the analytics may be helpful in providing the social benefits with the culturally acceptable practices but with the exposure of any individual’s orientation and medical history may lead to the discrimination towards those specific individual (Vayena et al., 2015).  For example, different researches revealed that countries/locations with tighter privacy regulations experience fewer privacy issues related to the analytics. Again with the greater control can have a downside and lead to lower advertising effectiveness for the organizations as well as consumer marketing outcomes.  It is also found that that in circumstances of higher perceived privacy control, consumers or the users are most likely to click on personalized ads generated using the analytics (Papernot & Goodfellow, 2018).

Impact on National Privacy Policies

With this technique it is possible to expose the identity of an individual through liking the different sources of data that in turn breaches the privacy of the individuals. 

Even though individuals contribute their information to the different organizations related to the different industries but ultimately the individuals do not have ownership over their data. The right of retaining one’s data implies that individuals should be benefited from   the contributed data. In this way the individuals using digital services from the organization such as google location based services or the services from the Facebook are often tracked by service providers against the usage of their services. In this way the users lose their privacy without their consent that leads to the ethical implication of use of the insights by the organizations.

Data analytics depends on different machine learning algorithms in order to support human decisions based on the huge amount of data.  The algorithmic decision-making takes input large set of data that are collected and combined from multiple data sources in order to predict some decision for the individual’s behaviour depending on their collected data about their past behaviour (Papernot & Goodfellow, 2018).  This decision making process focuses on identification of relationships in the different attributes of the collected data. In case of the bias in the facial recognition system it is observed that, the error rates are higher while detecting the darker skinned individual.

Again most of the machine learning systems/algorithms requires continuous input in order to train, improve and enhance the accuracy of the results of their algorithms. This process can be analogized to the process of quality improvement researches.  In this cases informed consent are considered as not necessary (Sankar & Parker, 2017). It is also desired to involve the stakeholders such as the citizens or the patients (in case of medical processes). Such as for research in the genetic diagnosis the algorithms require inputs the or images of individuals who have specific genetic disorders to improve the accuracy of  the results. In order to maintain trust as well as transparency with all the stake holders (in this scenario patients), organizations should be involving applicable community stakeholders in implementation process of FRT with their consent. Furthermore, Ethical issues arise at every phase of the processing and value creation through the analytics (Vayena et al., 2015). At the end of the process final owner of processed data and insights available from it may use the data insights for numerous purposes that are very different from the initial one (for which the data collected at the very first place).

Ethical Implications of Data Analytics

Furthermore, the ethical impacts of the biased use of the data analytics can be multifaceted.  Such an example can be stated as depending on the analysis any service provided by an organization can use filtered approach in order to hide pieces of information from different segment of users. In this way, the organization imposes a bias of which the users are unaware.

Again with the use of facial recognition systems it is possible for the organization to monitor and carry out surveillance on the individual trough the analysis of big data sets. Through the analysis of the behaviours can offer personalized services/ products (Sankar & Parker, 2017). At this stage this stated that these monitored individuals would not have exposure to all the available options/choices available in market places.

Ultimately, this leads to the fact that individuals are no longer subject to the basic rights related free choice for any product services.  In this scenario they get under the control of the surveillance and the algorithms that are developed in order to influence the decisions   of the individuals (Hajian, Bonchi & Castillo 2016).  In this way total process enforces the individuals under the surveillance capitalism with the impact of new drivers that are resulted from the insights available from big data insights rather than the traditional market-based capitalism.

In case of the big data analytics due to monetization of the available data of the users by the organizations without the consent of the owner or the users (Vayena et al., 2015). With the available insights from the data (which are not collected from the users for the analysis) the organizations are conducting analytics with ease and huge scale for influencing the individuals behaviour.  

Conclusion  

The insights from the analytics ay consist of considerable biases or errors as all the individuals not conforming to the same characteristics of a certain group. Additionally, when implications are made from the different data sets for specific individual may lead to poor quality. With the used algorithms, result may lead to the complex ethical issues. Both the government and private organizations are presently capable of combining diverse and multiple digital datasets available publically. Using this data sources, they can analyse them in order to find out the statistics and extract hidden insights/information and surprising correlation among them. This insight can lead to exploitation of the privacy   of the users about whom the data set contains information.   The decision making process assumes that detected relationships are meaningful in the light of the cause and effect scenario in a social phenomenon. One such example is to predict the matches of the human faces based on correlations in a collected and processed data set, which overlooks small precisions and inherent error and biases present in the collected data.

References 

Bradlow, E. T., Gangwar, M., Kopalle, P., & Voleti, S. (2017). The role of big data and predictive analytics in retailing. Journal of Retailing, 93(1), 79-95.

Dimitrov, D. V. (2016). Medical internet of things and big data in healthcare. Healthcare informatics research, 22(3), 156-163.

Hajian, S., Bonchi, F., & Castillo, C. (2016, August). Algorithmic bias: From discrimination discovery to fairness-aware data mining. In Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining (pp. 2125-2126). ACM.

Martin, K. D., & Murphy, P. E. (2017). The role of data privacy in marketing. Journal of the Academy of Marketing Science, 45(2), 135-155.

Martin, K. E. (2015). Ethical issues in the big data industry. MIS Quarterly Executive, 14, 2.

Papernot, N., & Goodfellow, I. (2018). Privacy and machine learning: two unexpected allies?.

Sankar, P. L., & Parker, L. S. (2017). The Precision Medicine Initiative’s All of Us Research Program: an agenda for research on its ethical, legal, and social issues. Genetics in Medicine, 19(7), 743.

Sivarajah, U., Kamal, M. M., Irani, Z., & Weerakkody, V. (2017). Critical analysis of Big Data challenges and analytical methods. Journal of Business Research, 70, 263-286.

Taylor, L. (2016). No place to hide? The ethics and analytics of tracking mobility using mobile phone data. Environment and Planning D: Society and Space, 34(2), 319-336.

Vayena, E., Salathé, M., Madoff, L. C., & Brownstein, J. S. (2015). Ethical challenges of big data in public health.

Vitak, J., Shilton, K., & Ashktorab, Z. (2016, February). Beyond the Belmont principles: Ethical challenges, practices, and beliefs in the online data research community. In Proceedings of the 19th ACM Conference on Computer-Supported Cooperative Work & Social Computing (pp. 941-953). ACM.