Encoding Sequential Information Using Signatures

摘要

This paper discusses two signature key related methods to encode sequence information with item information within a list. Using these methods will reduce the storage requirements for solving many-to-many relationships in very large databases (VLDB). The approach will reduce overall database sizes in many cases. In addition, with proper database configuration and tuning, system performance may improve for data linkage related reporting and/or analysis. These methods have an additional domain of application within pattern encoding and matching. With the proper decoding mechanism, partial pattern matching is also possible. By combining these methods with other techniques such as neural network and multi-set data type, they can further reduce storage and improve performance.