For the question how to apply the Web Information Extraction (WIE) technology to health field, a Web information extraction method based on WebHarvest was proposed. Through the structure analysis of different health Web sites and the design of health entity extraction rules, the automatic extraction algorithm of health entity and its attributes based on WebHarvest was realized; then they were stored in a relational database after consistency check; in the end, the values of entity attributes were analyzed to recognize entities by using processing method of natural language Ansj to extract relationship among entities. In the health entity extraction experiments, the average F-measure of the technology reached 99.9%; in the entity contact extraction experiments, the average F-measure reached 80.51%. The experimental results show that the proposed Web information extraction technology has high quality and credibility in the health information extraction.

Based on Shamir's threshold secret sharing scheme, the security of the linear one-way function and the difficulty of the discrete logarithm problem, a verifiable multi-secret sharing scheme was proposed．In this scheme, each participant needed just one secret share to share a set of secrets．Before recovering the secrets, participants could verify the correctness of the shadow shares provided by other participants．After recovering all of the secrets, the secret shares of the participants were still kept confidential and the secret shares could be used to share a new set of secrets．At the same time，the proposed scheme had fewer public parameters，and it did not require secure communication channels．

The Chase2 algorithm is one of the soft-decision decoding algorithms of Turbo Product Code (TPC). The conventional Chase2 algorithm needs a large amount of computation to calculate the Euclidean distance and search for competing codeword, so it is complex to implement in the hardware. Therefore, on the basis of the conventional Chase2 algorithm, the correlation was substituted as the metric for the Euclidean distance and the process of searching for competing codeword was simplified to reduce the decoding complexity; and the soft-output value was adjusted when there was no competing codeword to improve the coding gain. The simulation results show that, compared with the conventional algorithm, the modified algorithm has better performance and faster decoding speed, and it is suitable for hardware implementation.