الفهرس 
AbstractOnly 14 pages are availabe for public view
 |  | from | 100 |  |  |
| | | from | 100 | | |
Abstract
Cloud computing has been envisioned as a recent trend in the field of IT where computing and data storage are done in a cloud data centers rather than personal portable PC’s. Ensuring information security is a top concern regarding cloud computing to protect the data stored in the cloud servers from attacks. Privacy is considered as one of the most important issues since individuals upload their data in cloud storage server’s and have no direct control over the cloud system that manage their data. Cryptography is used to secure sensitive cloud data. It allows the users to encrypt the data before it is sent to the cloud for storage or processing and securely access various shared cloud services. Searching over encrypted data may be known as one of the most vital problems in cloud computing. Although the encryption process preserves data privacy, it undermines the efficiency of the search process especially on large scale data. Most of the existing search methods mainly focus on keyword-based search, which suffers from neglecting the semantic relations of the query keywords. This lack of semantic search capabilities negatively affects the efficiency of the search process and the search results may not completely match the user intent. This thesis introduces a semantic search scheme over encrypted data on the cloud in an efficient and accurate manner. The Latent Dirichlet Allocation (LDA) topic modeling algorithm is employed to cluster the data into similar coherent groups based on uncovering the hidden semantic structure of the data collection. The job of topic modeling is to limit the search space to some specific topics. So, enhance the search efficiency by reducing the search time. In addition, we utilize a method to expand the search query keywords based on their semantic relations to improve data retrieval effectiveness. Thus, the search results containing not only exact match documents but also semantically related documents to a search query. Finally, the search hits have been delivered to the search user in a ranked order. Experiments were conducted on real world dataset known as Request For Comments (RFC) dataset. The inclusive performance evaluation of the framework comprises the cost of topic modeling, the index construction time, the time required to retrieve the hits as well as the search efficiency. Results evaluation shows that the proposed method provides a powerful and secure search mechanism which is employed to index, search, rank, and retrieve encrypted documents from the cloud storage servers.