| 摘要 |
<p><P>PROBLEM TO BE SOLVED: To extract and output a pattern of high unexpectedness from state sequence patterns. <P>SOLUTION: State sequence patterns are created and registered in a state sequence pattern database 7. An in-database frequency p(s) calculation part 2 computes ratios of the occurrence frequency of a state sequence pattern s to the occurrence frequencies of "elements t of T[s]" of pattern subsets T[s] of the state sequence pattern s. About the state sequence pattern s, an assumed frequency calculation part 3 calculates assumed frequencies q(s) as conditional probabilities of the occurrence of s under the occurrence of the "elements t of T[s]" in a given probability distribution. An unexpectedness measure calculation part 4 computes measures I(s) of unexpectedness according to the in-database frequencies p(s), the assumed frequencies q(s), and relative occurrence frequencies of the "elements t of T[s]" in the pattern subsets T[s]. The results are evaluated in an evaluation part 5, processed in an output processing part 6 and output from an output device 9. <P>COPYRIGHT: (C)2004,JPO</p> |
