Research On An Approach Of Privacy Preserving Schema Matching

With the rapid development of information technology, how to efficiently access and use information is a significant problem faced by many business areas. Such issue has been resolved to some extent with the emergence of data integration and sharing. The key problem of data integration is to find the correct semantic mapping, which is schema matching. Schema matching as the first step, e.g. data warehousing, data integration, information sharing and network exchanging, plays a key role in many application areas.However, for the privacy considerations, people often do not easily send their information to the other party, only if the privacy of one's own is fully protected. Therefore, under the premise of how to protect privacy data and prevent sensitive information leakage, schema matching has become the major challenges.Most of the existing privacy protection techniques are aimed at privacy leakage in data mining process. This paper proposed a PPSM (Privacy Preserving Schema Matching) algorithm for schema matching process:first extract several of the most useful data norms from the norms which are used for describing the attributes. Classify the attributes according the data type of the attribute, and then we can get four attribute sets which have different types. Then convert the attribute name of mode information to the value type. Next, use four data norm matrixes to describe the four attribute sets. After that, convert the data norm matrixes and pass the converted data norm matrixes to the third party W. The schema matching result could be got according to the similarity of the data norm matrixes which are calculated by the third party W.Finally, in this thesis, it is verified that PPSM algorithm has a higher privacy protection by the experiment. And through classification, it improves the precision and comprehensiveness.