基于规则排序的介词用法自动识别研究

Automatic Annotation Research on Preposition Usage Based on Sorting Rules

在基于规则介词用法自动识别中,使用规则的不同顺序可以得到不同的用法识别准确率.对规则进行全排序并进行用法自动识别可以得到用法识别准确率最高的最佳排序方案,但是时间复杂度较高.首先对全排序的结果进行筛选,然后用筛选之后的结果进行用法自动识别则可以大大降低时间复杂度.实验结果表明,这种方法可以以较小的代价找到接近最佳排序的方案.

By using different sequences of rules in the automatic identification process of prepositional usage we could get different recognition accuracy.Having a full permutation of the rules firstly and then identi fying the sorting results automatically,we can get the best sequencing scheme by which the recognition accuracy rate would achieve the highest,while the time complexity is very high.However,screening the full permutation firstly and then taking automatic recognition upon the results can significantly reduce the time comple xity.The experimental results show that the latter method can be used as a shortcut to find the closest optimal sorting scheme with lower costs and thus obtain a higher identification accuracy of prepositional usage.