Since the advent of cloud computing, a huge number of data owners are outsourcing their sensitive data to be stored on to the cloud. To maintain privacy requirements, this sensitive data should be encrypted when it is stored on the cloud server which renders simple keyword based document retrieval schemes obsolete. In this paper, we propose a semantic multi-keyword ranked search scheme for document retrieval on cloud data that is encrypted. The proposed scheme returns not only the documents containing terms that match with our query terms but also some more documents that contain terms which are semantically similar to the query keywords. Our experimental result shows that our technique is more precise than TF-IDF/VSM(Vector Space Model) models that focus only keyword matching while simultaneously achieving faster retrieval times than other tree based TF-IDF/VSM approach as our algorithm runs on reduced dimensions. © 2020 IEEE.