,
The continuous evolution and pervasive application of communication technology, network technology, and computing technology have boosted interconnection of everything and ubiquitous sharing of information. With the continuous emergence of new information services, user data is frequently exchanged across systems, ecologies, and countries. User data contains a large amount of personal privacy information, which is retained intentionally or unintentionally in different information systems. At the same time, the data protection capabilities and protection strategies of each information system are very different, which leads to cask principle and a dramatic increase in the risk of privacy leakage. Protection of personal information and the governance of privacy abuse have become a worldwide problem.
Privacy preservation has received more and more attention from the society and extensive academic research, privacy preservation technologies for different scenarios are also emerging in spurts, and there are still many misunderstandings and confusion about the concept of privacy preservation. In particular, data security and privacy preservation are easily confused. This chapter will explain the connection and difference between user data, personal information, and privacy information, clarify the difference between data protection, privacy protection, and privacy desensitization, and point out the threats and technical challenges faced by privacy preservation.
F. Li (✉) · B. Niu
Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China e-mail: lifenghua@iie.ac.cn; niuben@iie.ac.cn
H. Li
School of Cyber Engineering, Xidian University, Xian, Shaanxi, China e-mail: lihui@mail.xidian.edu.cn
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 F. Li et al., Privacy Computing, https://doi.org/10.1007/978-981-99-4943-4_1
1.1 User Data, Personal Information, and Privacy Information
Privacy preservation should cover the full life cycle protection of privacy information. If we want the whole society to pay high attention to privacy preservation and put it into practice, the connotation of personal information and privacy information should be clarified, as well as the connection and difference between user data, personal information, and privacy information.
1.1.1 User Data
Data usually refers to a sequence of one or more symbols. Data can be observed, collected, processed, and analyzed, and through interpretation and analysis, data becomes information. From the perspective of information theory, data is the carrier of information. Data can be organized into many different types or data structures, such as lists, graphs, and objects. Data also has multiple modalities, such as numbers, text, images, video, and speech. Multimodal data can be exchanged across borders, systems, and ecosystems in a ubiquitous network environment.
User data can be data related to individuals, as well as related to businesses, organizations, objects, environments, and the like. In the era of the intelligent interconnection of everything and ubiquitous sharing of information, data has become a strategic resource, which is crucial to the interests and security of individuals, enterprises, society, and even the country.
1.1.2 Personal Information
As defined in the “Civil Code of the People’s Republic” of China, personal information is all kinds of information recorded electronically or in other ways that can identify a speci fic natural person alone or in combination with other information, including the natural person’s name, date of birth, identification (ID) number, biometric information, address, phone number, email, health information, whereabouts information, etc.
In Europe and North America, personal information mostly refers to personal data or Personally Identifiable Information (PII). Personal data is defined in the European Union’s General Data Protection Regulation (GDPR) [1] as any information relating to an identified or identifiable natural person (“data subject”); an identifiable natural person is one who can be identified, directly or indirectly, in particular by reference to an identifier such as a name, an identification number, location data, an online identifier or to one or more factors speci fic to the physical, physiological, genetic, mental, economic, cultural, or social identity of that natural person.
Table 1.1 Classification of personal information by the US Federal Trade Commission
Data elements
Identification data
Sensitive identification data
Demographic data
Court and public record data
Social media and technology data
Home and neighborhood data
General interest data
Financial data
Vehicle data
Travel data
Purchase behavior data
Health data
Data segments
Name, address(including latitude and longitude information), etc. 15 items
Social security number, driver ’s license number, etc. 5 items
Age, height, gender, religion, language etc. 29 items
Bankruptcy information, criminal information, judgments, etc. 7 items
Electronics purchases, friend connections, internet connection type, internet provider, social media and internet accounts, etc. 18 items
Census tract data, dwelling type, heating and cooling, home equity, move date, etc.24 items
Apparel preferences, preferred sports, favorite stars, political inclinations, etc. 43 items
Purchasing power, credit history, loan information, disposable income, etc. 21 items
Brand preference, vehicle identification code, preferred model, etc. 14 items
Last travel time, preferred destination, preferred airline, etc. 9 items
Purchase types, purchase channels, holiday gifts, and clothing sizes purchased, etc. 29 items
Search tendency for illnesses and medicines, smoking status, allergy information, etc. 15 items
The term PII is generally adopted by the United States. The U.S. Federal Trade Commission categorizes data related to natural persons, and divides personal information into 12 categories and 221 attribute fields. See Table 1.1 for speci fic categories [2].
Data records of personal information contain different fields, which can be divided into explicit identifiers, quasi-identifiers, sensitive attributes, and non-sensitive attributes. Explicit identifiers are sets of attributes that can clearly identify the identity of the record subject, including names, social security numbers, phone numbers, ID numbers, and other information. A quasi-identi fier is a collection of attributes that, when combined, can potentially identify the subject of a record, including information such as age, gender, zip code, and more. Sensitive attributes contain sensitive individual-specific information such as illness and salary. Note that the non-sensitive properties are all other properties that are not in the above three categories and the sets of these four types of fields are disjoint.
In the process of information service, personal information may exist explicitly in structured records, such as medical records in hospitals, student registration information in schools, household registration information in government departments, and vehicle and driver information in traf fic management departments. It may also exist in unstructured data such as Twitter, Moments, and pictures shared by many social networks. Identifying, measuring, and protecting users’ privacy information for different types of data records is an extremely complex and challenging problem.
1.1.3 Privacy Information
The “Civil Code of the People’s Republic of China” defines privacy as the tranquility of the private life of a natural person and the private space, private activities and private information that no one else should know about. Privacy information refers to sensitive information in personal information, which is a collection of identifiers, quasi-identi fiers, and sensitive attributes in personal information records. Privacy reflects the relationship between identifiers, quasi-identifiers, and sensitive attributes.
Privacy information is not static, and they have two typical characteristics: relative stability in a certain period of time and space-time dynamics. The dynamism means that privacy information usually changes with the changes of subjective preferences of natural persons, time, and scene. For example, some people are willing to publish text, photos, and other information that reflect their personal preferences on social networks, thinking that these are not privacy information, so the dynamic nature of privacy information is also subjective. Spatiotemporal dynamics bring greater technical challenges to privacy preservation.
1.2 Privacy Protection and Privacy Desensitization
To promote theoretical and technical research on privacy preservation, it is necessary to clarify the connection and distinction between traditional data security and privacy protection. Data security refers to ensuring the confidentiality, integrity, non-repudiation, and availability of data, mostly using technologies such as cryptography and access control. Data security technology ensures that the protected data is recoverable, that is, the information is lossless. User data contains privacy information, so data security techniques can naturally be applied to privacy protection. This book classifies such techniques as privacy protection. On the other hand, privacy preservation should make privacy information partially available in a ubiquitous interconnected environment while being protected. It is essential to achieve a balance between the strength of desensitization and the usability of information, which is the core connotation of privacy preservation, and also the new theory and technology needed for privacy preservation.
This book divides privacy preservation technologies into two categories: privacy protection and privacy desensitization. The privacy information protected by privacy protection technology is undistorted and reversible; the privacy information protected by privacy desensitization technology is distorted and irreversible. The evolution process of privacy preservation technology is shown in Fig. 1.1.
In 1982, Yao first proposed the secure bipartite computing protocol; Subsequently, Goldreich proposed a secure multi-party computing protocol
In 1978, Rivest, Adleman and Dertouzos found that RSA has some homomorphic characteristics and proposed the concept of homomorphic encryption; In 2009, Gentry constructed the first fully homomorphic encryption scheme
In 1998, Samarati and Sweeney proposed k-anonymity
In 2006, Machanavaijhala proposed l-diversity
In 2002, Rakesh et al. proposed an access control model for privacy preservation
In 2007, Li et al. proposed t-closeness.
In 2016, Li Fenghua et al. proposed a Cyberspace-oriented access control model
In 2006, Cynthia et al. proposed differential privacy
In 2013, Miguel proposed GeoIndistinguishability
In 2013, John et al. proposed local differential privacy
In 2015, Li Fenghua et al.proposed the theory and technology system of privacy computing
1.2.1 Privacy Protection
Privacy protection technology refers to the use of encryption, secure computing, access control, and other technologies to protect privacy information from unauthorized access, and the protected privacy information is reversible.
1.2.1.1 Encryption
Encryption is the most commonly used privacy protection technology, where personal information is encrypted for transmission, storage, and sharing. The encrypted information can only be decrypted and accessed with the decryption key. Although encryption protects the security of data, the encrypted data cannot be directly counted, handled, and processed, which will increase the complexity of using data. For encrypted data processing, two technology routes that are currently receiving wide attention from academia and industry are homomorphic encryption and confidential computing based on trusted computing environments. Homomorphic encryption means to perform function computation f(E(x)) on the ciphertext, and the result of decryption is equivalent to performing the corresponding function computation f(x) on the plaintext x x, that is, the encryption function E(x) and the function computation f(x) can exchange the order, i.e., D( f(E(x))) = f(D(E(x))) = f(x) D( f(E(x))) = f(D(E(x))) = f(x). With the support of homomorphic encryption, the user can encrypt the data and then hand it over to cloud computing or other partners. After the partner performs corresponding operations on the ciphertext, the user decrypts the ciphertext to obtain the computation result of the plaintext. RSA [3] and Pailiar algorithm [4] have the properties of multiplication and addition homomorphism, respectively, but general computing needs to have
Fig. 1.1 Evolution process of privacy protection technology
homomorphic properties for addition and multiplication at the same time. In 2009, Gentry [5] proposed the first fully homomorphic algorithm, which attracted widespread attention and inspired a lot of follow-up research. However, the complexity of the current fully homomorphic algorithm is still very high, and there still exists a large gap from practical application.
Confidential computing based on a Trusted Execution Environment (TEE) focuses on data protection in the computing process. The system maintains a secure space, decrypts the encrypted data after importing it into the secure memory space, calculates the plaintext, and encrypts it when it is called out of the space. This secure memory space is inaccessible to other users, reducing the risk of data leaking from other parts of the system while maintaining transparency to users. Especially in multi-tenant public cloud environments, confidential computing keeps sensitive data isolated from other authorized parts of the system stack. Intel SGX (Software Guard Extension) is currently the main method for implementing confidential computing, which generates an isolated environment Enclave in memory. SGX uses strong encryption and hardware-level isolation to ensure the confidentiality of data and code against attacks, and can still protect applications and code even when the operating system and BIOS firmware are compromised.
1.2.1.2 Secure Multiparty Computation
Secure Multi-Party Computation (MPC) originated from the secure two-party computation protocol “Millionaire Problem” proposed by Yao [6]. Computational parties cooperate to complete a computing problem without revealing their own sensitive information. As research progresses, there are already several practical cases for secure multi-party computation. The Boston Women’s Workforce Council used MPC in 2017 to calculate compensation statistics for 166,705 employees at 114 companies [7]. The company does not provide its raw data due to privacy concerns, and calculations show that the gender gap in the Boston area is even wider than previously estimated by the U.S. Bureau of Labor Statistics. To calculate the exact conversion rate from an ad to an actual purchase, Google calculated the size of the intersection between the list of people who viewed an ad for an item online and the list of people who actually bought the item. To calculate this value without exposing the list-specific data, Google uses a private intersection-sum protocol [8]. Although the protocol efficiency is not ideal, it is simple and can meet Google’s computing requirements.
1.2.1.3 Access Control
Access control is one of the most important approaches to achieving privacy preservation. The essence of privacy preservation is to share privacy information with authorized entities at the right time and in the right way. In traditional access control systems, permissions are formulated and implemented by system
administrators. Common access control strategies include discretionary access control, mandatory access control, and role-based access control. In privacy preservation scenarios, permissions and access control policies are basically set by the data owner. In application environments such as social networks and Internet services, privacy information is often forwarded by friends and spread across systems and ecosystems among different service providers. Therefore, extended control has become the biggest problem faced in privacy preservation scenarios. In 2016, Li Fenghua et al. [9] proposed a Cyberspace-oriented access control model and an extended control model.
Encryption can also be combined with access control. Attribute Based Encryption (ABE) is an encryption method that effectively implements access control [10] where users have several attributes, and each attribute is assigned a public-private key pair. When encrypting a plaintext, the encryptor selects the public key of the corresponding attribute to construct an encryption key according to the access control policy. This encryption key can directly encrypt the plaintext, or encrypt the key used to encrypt the plaintext. If the user has a private key with attributes that match the access control policy, the private key of the corresponding attribute is selected to construct the decryption key, and the corresponding ciphertext can be decrypted similarly. ABE is essentially a public key encryption system with relatively slow encryption and decryption speed.
1.2.2 Privacy Desensitization
Privacy desensitization protects privacy information by adopting a distorted and irreversible method, so that the desensitized information cannot be associated with the data subject. Privacy desensitization of the privacy information contained in the data includes but is not limited to existing methods such as Generalization, Suppression, Anatomization, Permutation, and Perturbation. New theoretical innovations in privacy desensitization are needed in the future. Privacy desensitization is also often referred to as privatization or anonymization.
1.2.2.1 Generalization
Generalization is a technique which replaces a speci fic value in a class of properties with a more general value. For example, if a person is 25 years old, it can be generalized to 20–30 years old; a person’s occupation is a programmer or a lawyer, and it can be generalized to white-collar workers (brain workers).
1.2.2.2
Suppression
Suppression refers to replacing an attribute, attribute value, or part of an attribute value with * when publishing information. For example, the mobile phone number is represented as 135****3675, and the credit card number is represented as 4392********.
1.2.2.3
Anatomization and Permutation
The goal of anatomization and permutation is to remove the association between quasi-identi fiers and sensitive attributes without changing the values of quasiidentifiers or sensitive attributes. Anatomization is to divide the original record table into two tables to publish, where one table publishes quasi-identi fier attributes, the other table publishes sensitive attributes, and the two tables only have the same GroupID as a common attribute. Permutation is used to divide a set of data records into groups, and permute sensitive values within the group, thereby disrupting the correspondence between quasi-identifiers and sensitive attributes.
1.2.2.4
Perturbation
Data perturbation refers to the technique of replacing the original data values with synthetic data values. Statistical information does not change signi ficantly after perturbation, and the perturbed data lose relevance to the real data subject. Classic data perturbation mechanisms include noise addition, data exchange, synthetic data generation, etc. Noise addition is mainly used for privacy preservation of numerical data by generating noise values from a speci fic distribution of noise and adding them to sensitive values. The main idea of data exchange is to exchange the values of sensitive attributes between personal data records, which can maintain low-order frequency statistics or marginal distributions for statistical analysis. Synthetic data generation aims of building a statistical model according to data, and then up-sampling from the model to replace the original data. Perturbation has a long history of application in statistical release control because of its simplicity, effectiveness, and the ability to maintain statistical information [11].
On the basis of the above desensitization operations, a series of privacy desensitization models and methods have been developed, including k-anonymity [12], ldiversity [13], t-closeness [14], differential privacy [15], local differential privacy [16], etc., which will be introduced in subsequent chapters.
1.3 The “Four Rights” of Privacy Preservation
GDPR has made relevant provisions on the right to know, the right to erasure, and the right to be forgotten. The right to know addresses the collection and processing of personal information, and the right to erasure and the right to be forgotten address the storage of personal information. With the popularization and application of mobile apps, although the right to know has not been fully implemented and has become the root cause of out-of-scope collection of privacy information, it has been widely noticed by everyone. In reality, data subjects voluntarily provide some privacy information to obtain personalized services, but the data subjects’ right to erasure and to be forgotten are privacy preservation issues that deserve more attention. Service providers’ neglect of the right to erasure and to be forgotten is a source of misuse of privacy information. Under the circumstance that privacy information is widely exchanged and disseminated in the ubiquitous interconnected environment, the “extended authorization” proposed by the authors of this book is the core criterion to ensure the controlled sharing of privacy information, and an effective guarantee mechanism for balancing privacy desensitization and usability.
1.3.1 Related Parties of Privacy Information
The related parties of privacy information are the participants in the processing of privacy information in the process of privacy preservation, including the following five aspects.
1. Data subject: refers to the owner of personal data or personal information.
2. Controller: refers to the natural or legal person, public authority, agency, or other body which, alone or jointly with others, determines the purposes and means of the processing of privacy information.
3. Processor: refers to a natural or legal person, public authority, agency, or other body which processes personal data on behalf of the controller.
4. Recipient: refers to a natural or legal person, public authority, agency, or another body, to which the privacy information are disclosed, whether a third party or not.
5. Third party: refers to a natural or legal person, public authority, agency, or body other than the data subject, controller, processor, and persons who, under the direct authority of the controller or processor, are authorized to process personal data.
1.3.2 Right to Know
The right to know requires the controller to obtain the consent of the data subject when collecting and processing personal information. The data subject has the right
to know how the data controller processes and stores personal information, where the personal information is obtained and to whom it will be transferred. When the purpose and method of processing personal information are changed, the individual’s consent shall be obtained again. When personal information processors transfer personal information to third parties, they should also inform individuals of the recipient’s identity and contact information. The recipient should continue to perform the obligations of the personal information processor. If there is any change, the data subject needs to be notified again and consent should be obtained.
1.3.3 Right to Erasure
The right to erasure is the right of the data subject to ask the controller to delete his or her personal information. When the data subject withdraws consent or the personal information is no longer necessary for the purpose for which it was collected and processed, the data controller can be asked to delete the relevant data. If the controller has made the data public, it should consider taking reasonable steps, including technical measures, to inform other data controllers who are processing personal data that the data subject has asked them to delete the personal information of the data subject, and the data controller and processor should delete personal information in a deterministic and irrecoverable manner.
1.3.4 Right to be Forgotten
The right to be forgotten means that when the storage period agreed between the data subject and the controller has expired or the processing purpose has been achieved, and the personal information controller or processor stops providing products or services, the controller or processor should take the initiative to delete personal information, i.e., personal information is automatically deleted after being retained by the data controller or processor for a certain period of time.
1.3.5 Extended Authorization
In social network applications, there are widespread problems such as personal information being forwarded more than twice by friends across Moments and systems. Therefore, in the process of dissemination of privacy information, whether the data subject can extend authorization for the cross-system exchange of personal information, and the implementation of extended control is crucial to privacy protection. The requirement for extended authorization is not mentioned in GDPR and other privacy protection-related regulations, but in the era of ubiquitous
information sharing, extended authorization is the basis for ensuring controlled sharing of privacy information. Extended control is a technical implementation method of extended authorization, and an indispensable and effective mechanism for balancing privacy desensitization and privacy information utility.
Although the “Personal Information Protection Law of the People’s Republic of China” requires that personal consent is required for personal information processing, if the purpose of processing personal information, processing methods, and types of personal information are changed, personal consent should be obtained again, but in in the actual information system implementation process, if there are no technical means to implement extended control mechanism, the legal requirements are difficult to implement.
1.4 Technical Challenges in Privacy Preservation
The development of technologies such as the Internet, mobile Internet, Internet of Things, cloud computing, 5G, and satellite communications has spawned an endless stream of new service paradigms, and privacy information flows widely across systems, ecosystems, and even across borders. From the perspective of the “threedimensional space” composed of time, scene, and privacy information, any privacy preservation scheme is a “point” in the three-dimensional space; It is necessary to form a privacy preservation algorithm system in the “three-dimensional space” that is continuous in time, universal in scenarios, and common to all privacy information modalities, so that the system can ensure the stability of the industry privacy information system and realize the protection of the full lifecycle, any scenario, and any privacy information. Although thousands of academic papers or solutions have been published on research in privacy preservation around the world, why hasn’t the problem of privacy preservation been effectively solved in practical applications? This is because privacy preservation has not theoretically solved a series of challenges in systematization, computability, etc.
1.4.1 Threats in Privacy Preservation
1.4.1.1 Over-Collection of Information
In recent years, with the rapid development of mobile Internet technology, mobile App has become the main carrier of network information service and an important tool for users to use mobile Internet. However, many mobile apps, while providing users with network services, have problems such as compulsory authorization, excessive claiming of rights, and frequent over-scope collection of personal information, leading to the increasingly serious threat of personal information leakage.
1.4.1.2 Unauthorized Use and Excessive Retention of Information in the
Information System
Network information service providers have collected a large amount of personal information, created user profiles without obtaining user consent and used “big dataenabled price discrimination”, illegally overstepping their authority to use personal information and illegally sharing personal information with third parties, resulting in excessive retention of personal information across systems and posing a huge risk of privacy leakage.
1.4.1.3
Inconsistency of Cross-System Protection Capabilities
Privacy information flows widely across systems, and the protection capabilities of each system may be inconsistent, so the protection capabilities of the whole life cycle of privacy information are limited by the system with the weakest protection capability. The lack of extended control mechanism for the flow of privacy information cross-system increases the risk of cross-system privacy information leakage.
1.4.2 Systematic Computational Model
In order to realize the full life cycle protection of privacy information and ensure its implementation in a ubiquitous information system, it is necessary to build a systematic computing model for the preservation of privacy information, which supports the measurement and on-demand protection of privacy information.
1.4.2.1
Perception and Dynamic Measurement of Privacy Information
Privacy information is multi-dimensionally correlated, scene-changing, and subjective, leading to dynamic changes in privacy cognition. To establish a systematic computing model, it is necessary to break through key bottlenecks such as the privacy information perception of diverse data, the fine-grained division of privacy attribute vectors, the dynamic quantification of privacy attributes, and the dynamic assessment of privacy information value and leakage risk associated with multiple factors, so as to solve the problem of accurate privacy perception in massive data and the time complexity problem of privacy dynamic measurement.
1.4.2.2 On-Demand Privacy Preservation and Combination of Privacy-Preserving Mechanisms
Regarding the changes in application scenarios, differences in privacy preservation requirements of data subjects, and diverse data types, it is necessary to study key technologies such as privacy desensitization mechanisms in the process of personal information collection, processing and sharing, privacy desensitization strategies adapted to scenarios, and parameter selection. At the meantime, regarding the different privacy preferences of users, diverse privacy preservation algorithms, and differences in protection degree requirements, etc., we should consider specific characteristics of different privacy algorithms, the relevance of multimodal data, and privacy preservation mechanisms, explore scenario-adapted methods for efficient and optimal combination of privacy preservation algorithms, and break through key technologies such as fine-grained feature description of privacy preservation algorithms, characterization of algorithm combination characteristics, detection, and resolution of multi-party privacy preference conflicts, to achieve automatic optimization and combination of multi-preserving algorithms for different types of data with differentiated privacy-preserving requirements.
1.4.3 Evaluation of Privacy-Preserving Effectiveness
Considering the diverse theoretical systems of different privacy-preserving algorithms, differences in application requirements and algorithm effects, etc., it is vital to study the multi-dimensional evaluation index system of privacy-preserving effectiveness, and propose a scenario-adapted quantitative evaluation model for effectiveness. It is necessary to break through key technologies such as qualitative and quantitative performance evaluation, privacy-preserving performance limit estimation of single algorithm and combined algorithm, and quantitative evaluation of privacy-preserving strength for dynamic data addition, etc., so as to provide multidimensional quantitative support for the evaluation of algorithm effectiveness and selection of algorithms.
Moreover, it is essential to study the evaluation method of privacy-preserving effectiveness based on big data analysis while taking into account the characteristics of big data in actual Internet applications, such as large time-scale, differentiated sources, large sample space, differences in the effect and continuous evolution of privacy-preserving algorithms, and correlation to privacy in desensitized information. It is necessary to break through key technologies such as accurate collection of cross-platform privacy-related background data, rapid perception and labeling of non-explicit privacy attributes, fine-grained data owner privacy knowledge modeling, multi-source scene content cross-correlated privacy leak detection, and Bayesian statistical inference based on mutual information between linked data, so as to
achieve a reverse evaluation of privacy-preserving effect based on privacy mining using big data analysis.
1.4.4 Extended Control of Privacy Information
The GDPR and other regulations do not explicitly mention the extended authorization of privacy, but there are widespread problems in social network applications such as personal information being forwarded more than twice by friends across Moments and systems. In view of the diverse social application scenarios, the differences in the sensitivity of the types of objects in the media, the diversity of data subjects, and the differences in the privacy-preserving capabilities of the platforms, considering the impact of cross-platform flowing and multiple forwarding of multi-subject data on privacy control, and impact of subject, object, social platform, privacy needs, and other elements on dissemination control, it is necessary to establish a fine-grained extended control mechanism that supports the random topology of social networks and the differential privacy-preserving requirements of multiple data subjects, and breaks through key technologies including the normalized description of multi-factor constraints such as scene, space-time, content, and privileges, secure binding of media with followed extended-control-policy of protection strength and constraint conditions, the sharing process monitoring based on labeling and exchange auditing, the dynamic logical relationship generation of the data cross-platform flowing, differential description of privacy-preserving strength of subject-related data objects, normalized description of data extended control policy across domains, so as to support the controlled sharing of data in ubiquitously connected environment.
1.4.5 Forensics of Privacy Infringement
To address the problems of dynamic and random dissemination paths of privacy information, hidden privacy infringement, and fragmented spatial and temporal distribution of evidence in the diversity of Internet applications, it is necessary to study infringement monitoring for multi-type data, whole-process infringement clue capture and analysis, and abnormal data sharing behavior judgment and traceability, break through key technologies such as cross-domain exchange control and violation determination, cross-platform multi-dimensional reconstruction of privacy infringement events, and virtual identity positioning, so as to achieve accurate tracking of privacy infringements.
1.5 Chapter Summary
In the continuous evolution of new business paradigms, frequent cross-border, crosssystem, and cross-ecosystem exchange of user data results in privacy information being retained intentionally or unintentionally in different information systems. The cask principle effect caused by the differences in the data protection capabilities and protection strategies of various information systems has led to an increasingly prominent risk of privacy leakage. Issues such as the lack of privacy information protection methods and the difficulty in governing the abuse of privacy information have become worldwide problems. Privacy preservation can be divided into two categories: privacy protection and privacy desensitization. Faced with the controlled sharing of information across systems, privacy desensitization technology is much needed to ensure the realization of the “four rights of privacy”. At present, many privacy preservation solutions have been proposed, which however cannot effectively address the privacy issues in practical applications. Therefore, it is a must to establish a new and complete theoretical system for the protection of privacy information in its entire life cycle to support privacy preservation in information systems. The notion of privacy computing proposed by Li Fenghua et al. [17] is the solution to the privacy preservation problem in practical applications, which will be elaborated in the subsequent chapters.
References
1. General data protection regulation. (2018)
2. Federal Trade Commission. Data brokers: a call for transparency and accountability. (2014)
3. Rivest, R.L., Adleman, L., Dertouzos, M.L.: On data banks and privacy homomorphisms. Foundations of Secure Computation. 4(11), 169–179 (1978)
4. Paillier, P.: Public-Key Cryptosystems Based on Composite Degree Residuosity Classes. In: International Conference on the Theory and Application of Cryptographic Techniques, pp. 223–238. Springer, Berlin (1999)
5. Gentry, C.: A fully homomorphic encryption scheme. Stanford University, Stanford (2009)
6. Yao, A.C.: Protocols for secure computations. In: IEEE 23rd Annual Symposium on Foundations of Computer Science, pp. 160–164. IEEE Press, Piscataway (1982)
7. Lapets, A., Jansen, F., Albab, K.D., et al.: Accessible privacy-preserving web-based data analysis for assessing and addressing economic inequalities. In: Proceedings of the 1st ACM SIGCAS Conference on Computing and Sustainable Societies, pp. 1–5. ACM Press, New York (2018)
8. Ion, M., Kreuter, B., Nergiz, E., et al.: Private intersection-sum protocol with applications to attributing aggregate ad conversions. IACR Cryptology ePrint Archivet. 7, 738 (2017)
9. Li, F., Wang, Y., Yin, L., et al.: Novel cyberspace-oriented access control model. J. Commun. 37(5), 9–20 (2016)
10. Sahai, A., Waters, B.: Fuzzy identity-based encryption. In: International Conference on Theory & Applications of Cryptographic Techniques, pp. 457–473. Springer, Berlin (2005)
11. Adam, N.R., Worthmann, J.C.: Security-control methods for statistical databases: a comparative study. ACM Comput. Surv. 12, 515–556 (1989)
12. Sweeney, L.: K-anonymity: a model for protecting privacy. Int J Uncertain Fuzz. 10(5), 557–570 (2002)
13. Machanavajjhala, A., Gehrke, J., Kifer, D., et al.: L-Diversity: Privacy beyond K-Anonymity. In: IEEE 22nd International Conference on Data Engineering, p. 24. IEEE Press, Piscataway (2006)
14. Li, N., Li, T., Venkatasubramanian, S.: T-Closeness: Privacy beyond K-Anonymity and L-Diversity. In: IEEE 23rd International Conference on Data Engineering, pp. 106–115. IEEE Press, Piscataway (2007)
15. Dwork, C.: Differential privacy: a survey of results. In: International Conference on Theory and Applications of Models of Computation, pp. 1–19. Springer, Berlin (2008)
16. Duchi, J.C., Jordan, M.I., Wainwright, M.J.: Local Privacy and Statistical Minimax Rates. In: IEEE 54th Annual Symposium on Foundations of Computer Science, pp. 429–438. IEEE Press, Piscataway (2013)
17. Li, F., Li, H., Jia, Y., et al.: Privacy computing: concept, connotation and its research trend. J. Commun. 37(4), 1–11 (2016)
Another random document with no related content on Scribd:
folded it, put it in its envelope, and fastened a look that a basilisk might have envied, on her companion.
Glancing up from her novel with a frank fearless countenance, she encountered Miss Fane’s cold gray eyes critically surveying her, over the top of her tortoiseshell pince-nez. To describe Miss Fane more particularly, she was a prim, dignified, elderly lady, seated bolt upright on the most uncompromising chair in the room. She had wellcut aristocratic features; a high arrogant-looking nose; rather a spiteful mouth; iron-gray sausage curls, carefully arranged on either temple, and surmounted by a sensibly sedate cap. A very handsome brown silk dress, as stiff as herself, completed her costume.
Not being overburdened with this world’s goods, owing to the failure of a bank in which most of her fortune had been invested, she had accepted a very handsome allowance and the post of chaperon to her nephew’s ward. If she could have had this immense increase to her income without the ward, so much the better; girls were not to her taste, but though narrow-minded, frigid, and intensely selfish, she was strictly conscientious, according to her lights, and was thoroughly prepared to do her duty by her young companion.
“Alice,” she said, glancing from Alice to the note she held in her hand, and then back again with an air of hesitation, “I have just heard from my nephew, your guardian, you know. He expects to leave India immediately; and if the Euphrates stops here for coaling, he says he will come and look us up. Would you like to read his letter? Perhaps I ought not to show it to you; but it will give you some idea of the kind of young man he is.”
“Thank you,” replied his ward, stretching out a slim ready hand; “if you really think I may, Miss Fane,” she added interrogatively, whereupon Miss Fane handed her her nephew’s effusion, which ran as follows:
“Cheetapore.
“M