If it is the author's pre-published version, changes introduced as a result of publishing processes such as copy-editing and formatting may not be reflected in this document. For a definitive version of this work, please refer to the published version.

Abstract. The traditional recommendation algorithms of image tagging ignore the diversity between the visual content information and the tags recommended, which causes the recommended results have the problem of tag ambiguity, tag redundancy and so on. Therefore, this paper proposes the recommendation algorithm of image tagging based on relevance and diversity. The algorithm defines the relevance and diversity of a label set, and selects a label set which can reasonably balance the relevance and diversity to recommend to the user. The experimental results show that this algorithm improves the relevance between the recommended results and the image, and makes the recommended results be able to reflect the image content thoroughly at the same time.

Abstract. How to protect sensitive data privacy is one of the research focus in the field of data mining, especially in the case of the data distributed storage, and the meaning of data privacy protection is particularly important. Secure multi-party computation security multiparty computation password primitives in the privacy of distributed data mining related applications, the literature [1] of privacy protection data mining model based on security multiparty computation were analyzed, and the argument of this model is based on discrete logarithm public-key encryption protocol is not fully homomorphic characteristics, with a simple example. Thus, it is concluded that the privacy data mining model is not feasible.

No abstract

The Internet has penetrated into every aspect of life. Large amounts of data are generated by multimedia collection equipment every day. As an asset, data can achieve value circulation through transactions. However, the existing centralized transaction model is not secure enough, has the risk of user privacy leakage, and the protection of data copyright is insufficient. In this paper, in order to solve the transaction security and traceability problems of multimedia data, especially high-definition data such as vector graphics, we implement a transaction scheme STTS without third-party based on blockchain. For high-definition multimedia data, we use zero watermarking combined with oblivious transfer to embed copyright information. A two-stage verification process is then implemented by using group signature and a secret sharing scheme to complete data distribution. Finally, the smart contract is used to complete copyright tracking. We test the performance of our scheme by simulating the real transaction environment in the Internet of Things (IoT) and demonstrate the feasibility of our scheme, which can be applied to large-scale multimedia data trading schemes in the IoT.