Privacy-Preserving Multi-Keyword Fuzzy Adjacency Search Strategy for Encrypted Graph in Cloud Environment

In a cloud environment,outsourced graph data is widely used in companies,enterprises,medical institutions,and so on.Data owners and users can save costs and improve efficiency by storing large amounts of graph data on cloud servers.Servers on cloud platforms usually have some subjective or objective attacks,which make the outsourced graph data in an insecure state.The issue of privacy data protection has become an important obstacle to data sharing and usage.How to query outsourcing graph data safely and effectively has become the focus of research.Adjacency query is a basic and frequently used operation in graph,and it will effectively promote the query range and query ability if multi-keyword fuzzy search can be supported at the same time.This work proposes to protect the privacy information of outsourcing graph data by encryption,mainly studies the problem of multi-keyword fuzzy adjacency query,and puts forward a solution.In our scheme,we use the Bloom filter and encryption mechanism to build a secure index and query token,and adjacency queries are implemented through indexes and query tokens on the cloud server.Our proposed scheme is proved by formal analysis,and the performance and effectiveness of the scheme are illustrated by experimental analysis.The research results of this work will provide solid theoretical and technical support for the further popularization and application of encrypted graph data processing technology.

An Efficient Multi-Keyword Query Processing Strategy on P2P Based Web Search

The paper presents a novel benefit based query processing strategy for efficient query routing. Based on DHT as the overlay network, it first applies Nash equilibrium to construct the optimal peer group based on the correlations of keywords and coverage and overlap of the peers to decrease the time cost, and then presents a two-layered architecture for query processing that utilizes Bloom filter as compact representation to reduce the bandwidth consumption. Extensive experiments conducted on a real world dataset have demonstrated that our approach obviously decreases the processing time, while improves the precision and recall as well.

Paillier-Based Fuzzy Multi-Keyword Searchable Encryption Scheme with Order-Preserving

Efficient multi-keyword fuzzy search over encrypted data is a desirable technology for data outsourcing in cloud storage.However,the current searchable encryption solutions still have deficiencies in search efficiency,accuracy and multiple data owner support.In this paper,we propose an encrypted data searching scheme that can support multiple keywords fuzzy search with order preserving(PMS).First,a new spelling correction algorithm-(Possibility-Levenshtein based Spelling Correction)is proposed to correct user input errors,so that fuzzy keywords input can be supported.Second,Paillier encryption is introduced to calculate encrypted relevance score of multiple keywords for order preserving.Then,a queue-based query method is also applied in this scheme to break the linkability between the query keywords and search results and protect the access pattern.Our proposed scheme achieves fuzzy matching without expanding the index table or sacrificing computational efficiency.The theoretical analysis and experiment results show that our scheme is secure,accurate,error-tolerant and very efficient.

Authorized Attribute-Based Encryption Multi-Keywords Search with Policy Updating

Attribute-based encryption is cryptographic techniques that provide flexible data access control to encrypted data content in cloud storage.Each trusted authority needs proper management and distribution of secret keys to the user’s to only authorized user’s attributes.However existing schemes cannot be applied multiple authority that supports only a single keywords search compare to multi keywords search high computational burden or inefficient attribute’s revocation.In this paper,a ciphertext policy attribute-based encryption(CP-ABE)scheme has been proposed which focuses on multi-keyword search and attribute revocation by new policy updating feathers under multiple authorities and central authority.The data owner encrypts the keywords index under the initial access policy.Moreover,this paper addresses further issues such as data access,search policy,and confidentiality against unauthorized users.Finally,we provide the correctness analysis,performance analysis and security proof for chosen keywords attack and search trapdoor in general group model using DBDH and DLIN assumption.

Research on Fuzzy Matching Query Algorithm Based on Spatial Multi-keyword

With the rapid growth of spatial data,POI(Point of Interest)is becoming ever more intensive,and the text description of each spatial point is also gradually increasing.The traditional query method can only address the problem that the text description is less and single keyword query.In view of this situation,the paper proposes an approximate matching algorithm to support spatial multi-keyword.The fuzzy matching algorithm is integrated into this algorithm,which not only supports multiple POI queries,but also supports fault tolerance of the query keywords.The simulation results demonstrate that the proposed algorithm can improve the accuracy and efficiency of query.

Dynamic multi-keyword fuzzy ranked search with leakage resilience over encrypted cloud data

To achieve the confidentiality and retrievability of outsourced data simultaneously,a dynamic multi-keyword fuzzy ranked search scheme(DMFRS)with leakage resilience over encrypted cloud data based on two-level index structure was proposed.The first level index adopts inverted index and orthogonal list,combined with 2-gram and location-sensitive Hashing(LSH)to realize a fuzzy match.The second level index achieves user search permission decision and search result ranking by combining coordinate matching with term frequency-inverse document frequency(TF-IDF).A verification token is generated within the results to verify the search results,which prevents the potential malicious tampering by cloud service providers(CSP).The semantic security of DMFRS is proved by the defined leakage function,and the performance is evaluated based on simulation experiments.The analysis results demonstrate that DMFRS gains certain advantages in security and performance against similar schemes,and it meets the needs of storage and privacy-preserving for outsourcing sensitive data.

基于联盟区块链和星际文件系统的安全租房方案

近年来租房市场蓬勃发展,各种租房平台应运而生。但现有的租房平台中心化问题严重,一旦中心服务器遭到破坏,容易引起单点失败问题,同时泄露用户的隐私。联盟区块链以分布式方式存储数据和执行合约,不依赖单个节点的可用性,即使某个中心节点失效,其他节点仍然可以继续运行和验证交易,确保系统的持续稳定性和安全性。提出一个基于联盟区块链和星际文件系统(IPFS)的安全租房方案。在方案中,房东通过预设访问策略对房源数据进行属性基加密,以实现数据的机密性和安全的细粒度访问控制。采用多关键词排序搜索,只返回最符合用户需求的前k个文件,从而提高搜索精度,同时节省网络带宽资源。最后,在区块链节点上部署搜索智能合约和验证智能合约用于陷门与索引的匹配以及搜索结果正确性的检验,同时为用户引入信誉智能合约和支付智能合约,前者用于量化信誉度,后者用于实现自动化执行租赁合约。安全性证明和性能分析表明,该方案达到了选择明文安全性及较高的搜索效率。

支持语义扩展的多关键词密文检索方案

现有的多关键词的检索研究集中在支持关键词的精确匹配,文中提出一个支持语义扩展查询的多关键词密文检索方案,基于信息内容计算方法同时考虑到关键词与文档的关系引入TF-IDF技术实现语义扩展;采用平衡二叉树的索引结构,使用安全k近邻算法加密索引向量和语义扩展查询向量,确保计算出准确的相关性得分,实现多关键词排序搜索。通过进行索引和陷门的机密性、查询无关联性等安全性分析表明,本文实现的检索方案是安全的,可以防止内部威胁。

基于字典分级和属性加权的密文排序检索方案

可搜索加密支持用户在不解密原始数据的前提下对加密数据执行检索操作.现有的多关键词排序可搜索加密方案,其索引和陷门构建的时间成本通常依赖于由全局关键词字典张成的向量空间.为了减少用户端的计算开销和通信成本,进一步提升数据使用者对检索结果的满意度,提出了一种支持细粒度访问控制的多关键词密文排序检索方案.该方案首先设计基于互信息的字典剥离机制差异化全局字典中的关键词,得到两个信息量不同的附属子字典,进一步在低维子字典空间上生成索引和陷门;其次,引入文档访问策略中属性的权重,将其作为排序标准之一,使数据使用者获得更为相关的结果;最后,检索时利用筛选向量对数据进行初次过滤并借助属性匹配完成二次剔除,从而避免检索过程中不必要的计算.

基于异构图和关键词的抽取式文本摘要模型

抽取式文本摘要使用一定的策略从冗长的文本中选择一些句子组成摘要,其关键在于要尽可能多地利用文本的语义信息和结构信息。为了更好地挖掘这些信息,进而利用它们指导摘要的抽取,提出了一种基于异构图和关键词的抽取式文本摘要模型(HGKSum)。该模型首先将文本建模为由句子节点和词语节点构成的异构图,在异构图上使用图注意力网络学习节点的特征,之后将关键词抽取任务作为文本摘要任务的辅助任务,使用多任务学习的方式进行训练,得到候选摘要,最后对候选摘要进行精炼以降低冗余度,得到最终摘要。在基准数据集上的对比实验表明,该模型性能优于基准模型,此外,消融实验也证明了引入异构节点和关键词的必要性。

基于区块链的动态密文排序检索方案

为解决云存储服务器不可信问题,提出基于区块链的动态密文排序检索方案。首先,采用树形索引结构,可达到亚线性的检索效率;使用向量空间模型,降低了文本的复杂度;其次,利用词频-逆文档频率(TF-IDF)加权统计算法实现了多关键词检索结果的排序;再次,采用为新添加的文件单独建立索引树、为删除文件建立删除列表的方法,动态更新区块链上可搜索加密方案;最后,通过泄漏函数证明了所提方案对自适应选择关键词攻击的安全性。性能测试结果表明,所提方案采用的树形索引结构相较于{key,value}索引结构,索引树生成时间平均降低了98%、文件搜索时间平均降低了7%、动态更新时间平均降低了99%,各阶段效率均有明显提高。

面向医疗系统的区块链的可搜索加密方案

在云计算作为辅助的电子医疗系统中,患者的电子医疗记录(Electronic Healthcare Records,EHRs)通常会外包给云服务器提供商(Cloud Server Provider,CSP),其中EHRs一般会以加密的形式上传到云服务器,再通过可搜索加密方案进行搜索.然而,由于过度依赖于被认为可完全信任的中心化服务器,现有的大多数可搜索加密方案仍面临着严重的安全问题.论文提出了一个面向医疗系统的区块链的可搜索加密方案,它不仅可以确保EHRs的安全,还可以提高存储在云服务器上的密码文本的搜索效率.在方案中,患者可以利用智能合约构建自动执行与自动查找的算法,这使医生收到可信的、正确的搜索结果.同时,方案采用了基于关键词转换的高效的模糊多关键词可搜索加密,优化EHRs的提取方式进而减少计算开销.此外,方案做了安全性分析和性能评估,证明方案的有效性和安全性.

基于区块链的公平可验证的多关键词密文排序检索

针对区块链环境下已有的可搜索加密方案实现结果验证和公平支付的成本过高、检索功能局限的问题,提出基于区块链的支持验证与公平支付的多关键词排序检索方案。该方案通过云服务器(CSP)存储加密索引树和执行搜索操作,并且构建了包含验证证明的查找表来辅助智能合约完成检索结果的验证以及公平支付,从而降低智能合约执行操作的复杂性,节约时间和费用成本。此外,结合向量空间模型与词频逆文档频率(TF-IDF)技术构建平衡二叉树结构的索引,并使用安全K邻近对索引和查询向量进行加密,从而实现支持动态更新的多关键词排序检索。安全性和性能分析表明,所提方案在区块链环境下和已知密文模型下是安全可行的;仿真实验结果表明,所提方案能够以可接受的开销实现结果验证与公平支付。

基于区块链的多关键词模糊搜索加密方案

针对1对多数据密文共享中多关键词模糊匹配和用户公平性问题,该文提出一种基于区块链的多关键词模糊搜索加密方案。该文提出一种R-HashMap索引结构,通过使用对偶编码函数和位置敏感哈希函数来构建安全索引,并采用K最近邻算法来加密索引,通过计算欧式距离度量查询关键词向量与索引节点之间的相似性,实现多关键词模糊密文搜索。该文除了消除预定义词典和降低存储开销外,还在不增加搜索复杂度的前提下实现对安全索引的更新。此外,将以太坊区块链技术与可搜索加密方案相结合避免了恶意服务器对数据的篡改,使用智能合约作为可信第三方进行检索工作,不仅可以防止云服务器内部的关键词猜测攻击,还可以解决检索结果不正确的问题。通过安全性证明分析,该文不但满足自适应选择关键词语义安全性,还可以保护用户隐私和数据安全。将该文与其他方案进行实验对比,证明该文在保证精确度的前提下,时间开销上具有更好的效率优势。

区块链下的EMRs聚合密钥可搜索加密方案

针对患者的多个电子医疗病历中单关键字检索的安全性不足和检测效率低的问题,提出一种区块链下的电子病历(EMRs)聚合密钥可搜索加密方案。采用星际文件系统(IPFS)和云服务器结合存储EMRs和密文,由患者在区块链上设置智能合约和存储交易单,对多文档多关键字的EMRs提出聚合密钥可搜索加密方案,数据请求方通过两次陷门匹配保证数据有针对性且安全的共享。安全性分析和数值模拟实验结果表明,该方案在保证聚合密钥安全和密文安全的前提下,具有较高的计算效率。

3-Multi ranked encryption with enhanced security in cloud computing

Searchable Encryption(SE)enables data owners to search remotely stored ciphertexts selectively.A practical model that is closest to real life should be able to handle search queries with multiple keywords and multiple data owners/users,and even return the top-k most relevant search results when requested.We refer to a model that satisfies all of the conditions a 3-multi ranked search model.However,SE schemes that have been proposed to date use fully trusted trapdoor generation centers,and several methods assume a secure connection between the data users and a trapdoor generation center.That is,they assume the trapdoor generation center is the only entity that can learn the information regarding queried keywords,but it will never attempt to use it in any other manner than that requested,which is impractical in real life.In this study,to enhance the security,we propose a new 3-multi ranked SE scheme that satisfies all conditions without these security assumptions.The proposed scheme uses randomized keywords to protect the interested keywords of users from both outside adversaries and the honest-but-curious trapdoor generation center,thereby preventing attackers from determining whether two different queries include the same keyword.Moreover,we develop a method for managing multiple encrypted keywords from every data owner,each encrypted with a different key.Our evaluation demonstrates that,despite the trade-off overhead that results from the weaker security assumption,the proposed scheme achieves reasonable performance compared to extant schemes,which implies that our scheme is practical and closest to real life.

面向云环境密文排序检索的字典划分向量空间模型

针对传统向量空间模型(TVSM)生成的向量维度高,计算文档与检索关键词相关度的向量点积运算耗时长的问题,提出一种面向云环境密文排序检索的字典划分向量空间模型(DPVSM)。首先给出DPVSM的具体定义,并证明了DPVSM中检索关键词与文档的相关度得分与TVSM中的相关度得分完全相等;然后,采用等长字典划分方法,提出加密向量生成算法和文档与检索关键词相关度得分计算算法。实验结果表明,DPVSM文档向量的空间开销远少于TVSM,且文档数量越多开销降低越多;此外,DPVSM的检索向量的空间开销以及相关度得分计算的耗时也远低于TVSM。显然,DPVSM在生成向量的空间效率和相关度得分计算的时间效率上均优于TVSM。

基于区块链的属性基多关键词排序搜索方案

对云数据进行访问控制能够限制非法访问、提高数据隐私安全,属性基可搜索加密是实现数据细粒度访问控制的关键技术之一。针对云数据访问中单一授权性能瓶颈、搜索功能局限等问题,提出一种基于区块链的属性基多关键词排序搜索方案。该方案采用多授权机制降低了系统计算负担,同时将属性基可搜索加密技术与区块链技术相结合,实现了云数据的细粒度访问控制与公平搜索;此外,引入向量空间模型和TF-IDF加权技术实现了多关键词搜索结果排序,提高了搜索效率。安全性分析、性能分析表明,该方案能够抵抗选择明文攻击和关键词猜测攻击,并具备较低的通信和计算开销。

基于可验证秘密共享的多关键词可搜索加密方案

可搜索加密技术在云存储技术快速发展的环境下已经被广泛应用于数据安全与个人信息保护的实施办法中,并作为研究热点被国内外众多学者关注。密码体制的安全性依赖密钥的保密,然而目前大多可搜索加密方案都需要用户自行管理密钥,存在密钥泄露和遗失的风险,影响加密方案的安全性,同时也限制了需要使用不同设备的用户。文章针对可搜索加密的密钥管理问题,基于可验证秘密共享提出了一种密钥可重构的多关键词可搜索加密方案,用户只需依赖自己的生物特征和预设口令即可完成数据的外包和检索,摆脱了自行管理密钥的困扰。方案满足选择关键词不可区分和身份认证两个安全属性,确保只有口令正确且生物特征与模板足够接近的情况下才能完成数据的外包和检索,用户特征的正确性由服务器端计算并验证,且无法得到关于用户的原始生物特征、密钥和关键词等隐私信息。

一种基于SWF-BERT的软件缺陷报告严重性预测方法

针对软件缺陷报告严重性预测中现有模型分类精度较低、深层次的语义特征不够丰富等问题,本文提出了一种基于BERT句子级别与词级别特征融合的SWF-BERT(Sentence-level and Word-level features Fusion-BERT)软件缺陷报告严重性预测模型.首先,对缺陷报告中的文本进行了数据预处理.其次,为了加强嵌入层中融合后的特征语义信息,提取词频最高的前100个单词,筛选出与缺陷严重性相关的特征词对其进行关键词嵌入操作,并融合嵌入层中的其他向量进行词嵌入.最后,将BERT模型输出层得到的特征(除[CLS]token外)送入多尺度卷积神经网络结合长短期记忆网络(MC-LSTM)模型中,加强了不同特征间远距离的时序信息.采用BERT模型输出得到的[CLS]句向量经过线性变换的结果与MC-LSTM模型输出经过线性变换得到的结果做可学习的自适应加权融合,实现了对软件缺陷报告严重性的有效预测.实验结果表明,使用SWF-BERT模型的平均准确率、召回率和F1值在Mozilla数据集中分别达到了68.41%、64.60%和64.86%,在Eclipse数据集中分别达到了61.32%、62.62%和59.31%,与其他分类算法相比,该方法在性能上得到了较大的提升.