| 摘要 |
PROBLEM TO BE SOLVED: To detect correlation of pieces of data from a combination of keys in the data to thereby group the pieces of data having the correlation.SOLUTION: Pieces of data a-j belonging to a segment S0 are locally divided to child segments S1, S2. A graph G0 is graphically divided to sub-graphs G1, G2. Keys ka-kj of the pieces of data a-j indicate nodes. In the sub-graphs G1, G2, sub-graph feature values C(G1), C(G2) are calculated. Key feature values are acquired in the pieces of data a-j in the segment S0. The pieces of data are respectively grouped into two: one pieces of data having a key feature value close to the sub-graph feature value C(G1); and the other pieces of data having a key feature value close to the sub-graph feature value C(G2). The segment S1 has the pieces of data a-d, f having the key feature value close to the sub-graph feature value C(G1), and the segment S2 has the pieces of data e, g-j having the key feature value close to the sub-graph feature value C(G2). |
