Attribute-based keyword search (ABKS) has been proposed to realize fine-grained access control and provide search service in cloud computing. However, most ABKS schemes focus on single or conjunctive keyword search, while the recent Boolean keyword search schemes only support monotonic query formula mainly involving AND, OR and threshold operators. How to support more expressive Boolean query formulas and return the corresponding accurate search results to users have become challenges for practical ABKS over ciphertexts. In this paper, we introduce an attribute-based expressive and ranked keyword search scheme over encrypted documents named ABERKS, which allows authorized users to submit expressive Boolean query formulas involving AND, OR, NOT and threshold operators. ABERKS utilizes a non-monotonic access tree structure to construct the query formula, and further leverages extended Boolean model to rank the search results. Specifically, the users are able to define the weights in the query formula, and get the relevance score of each matched ciphertext if the attributes and keywords are both satisfied. We prove the security of ABERKS against chosen keyword attack under selective ciphertext policy model and against keyword guessing attack, and also conduct extensive experiments to show the efficiency and practicality of ABERKS.

中文翻译：

---

云计算中加密文档的基于属性的表达和排序关键字搜索

基于属性的关键字搜索（ABKS）被提出来实现细粒度的访问控制并在云计算中提供搜索服务。然而，大多数 ABKS 方案侧重于单个或联合关键字搜索，而最近的布尔关键字搜索方案仅支持单调查询公式，主要涉及 AND、OR 和阈值运算符。如何支持更具表现力的布尔查询公式，并将相应的准确搜索结果返回给用户，成为实用ABKS over 密文面临的挑战。在本文中，我们介绍了一种名为 ABERKS 的加密文档的基于属性的表达和排序关键字搜索方案，它允许授权用户提交涉及 AND、OR、NOT 和阈值运算符的表达布尔查询公式。ABERKS 利用非单调访问树结构构造查询公式，并进一步利用扩展布尔模型对搜索结果进行排序。具体来说，用户可以在查询公式中定义权重，在属性和关键词都满足的情况下，得到每条匹配密文的相关性得分。我们证明了ABERKS在选择性密文策略模型下抵抗选择关键字攻击和抵抗关键字猜测攻击的安全性，并进行了大量实验以证明ABERKS的有效性和实用性。