LANGUAGE ANALYSIS DEVICE, METHOD, AND PROGRAM

摘要

In a case where the maximum input length which an analysis process can accept is given, it is impossible to divide a long sentence into a unit of processing which has a length appropriate to the maximum input length. A language analysis device includes a division point candidate generating means (21), a division point adjustment means (22), and a division means (23). In a case where a division rule which is classified in accordance with the risk of causing a problem in the analysis accuracy by applying the rule and a character string of which the length is longer than the maximum input length which is determined in advance are inputted, the division point candidate generating means (21) generates a division point candidate for the inputted character string by sequentially applying the division rules in the ascending order of levels of the risk of causing a problem. In a case where a division unit candidate obtained from the division point candidate which is generated by the division point candidate generating means (21) is shorter than the maximum input length, the division point adjustment means (22) selects a combination of division points out of the division point candidates obtained in accordance with the division rules of the same level so that each unit of division should not exceed the maximum input length. The division means (23) divides the inputted character string at the division point determined by the division point adjustment means.