| 摘要 |
PROBLEM TO BE SOLVED: To provide a context generation apparatus capable of providing context closer to context of target data.SOLUTION: The context generation apparatus includes: a previous probability estimation section for calculating a first model formed from occurrence probabilities of data values in data for learning; a context estimation section for calculating a second model formed from occurrence probabilities of the data values in the case where a reference data value is given, in the data for learning; a mixture ratio determination section which refers to the second model to determine the first model, a predetermined candidate model, the candidate model being formed from the occurrence probabilities of the data values in the case where the reference data value is given, and a mixture ratio with the first model; and a context model generation section which generates a context model by mixing the first model and the candidate model in the mixture ratio. |
