Study on linked data compression based on vector representation

In recent years, with the crazy growth of Internet data, the linked data also increased sharply. In such an Internet environment, storing and transmitting these data requires a huge disk and bandwidth overhead. Therefore, it is crucial to find an efficient RDF compression scheme for data storage and transmission. The current data compression technology addresses the syntactic and semantic redundancy of RDF. However, structured and repetitive redundancy has not been effectively addressed. This paper presents a scheme of RDF compression based on predicate index and vector. First, Map the original triples to ID-triples to reduce the length and redundancy of the statements. Second, clustering triples according to the predicate and use the vector to represent clusters, reducing the redundant information of linked data and improving the compression ratio. Then, clustering the triples according to the subject and the object, and construct SP index storage and OP index storage with variable length coding. In the case of the subject or object is known, the corresponding predicate can be quickly located and the speed of the query is improving. Experiments show that this compression scheme is quite competitive in the existing compression methods.