Data clustering method and device

Abstract

The invention provides a data clustering method and device in order to solve the technical problem of poor clustering effect caused by influence of initial conditions in an existing question set clustering process. The data clustering method comprises steps as follows: to-be-processed data is acquired and comprises testing data and non-testing data; the testing data is subjected to first classification processing, and a first classification result is obtained; the testing data is subjected to second classification processing by adopting an initial preset value, and a second classification result is obtained; the second classification result and the first classification result are compared, and when the accuracy rate of the second classification result obtained by utilizing the first classification result as the standard is larger than or equal to a threshold value, the initial preset value is set as a target preset value; when the accuracy rate is smaller than the threshold value, the initial preset value is adjusted constantly until the accuracy rate of a new second classification result obtained when the initial preset value is adjusted to be the target preset value is larger than or equal to the threshold value; the non-testing data is subjected to the second classification processing by adopting the target preset value.

Description

technical field [0001] The invention relates to a data processing method and device, in particular to a corpus data processing method and device. Background technique [0002] In the field of automatic question answering of language processing, it is necessary to determine the questions with language as the carrier, and then establish the corresponding relationship between questions and answers, and establish a question set of similar questions, that is, the aggregation of question sets is to determine the "question-answer" business logic The basic technology and important steps. [0003] In the process of aggregation processing of question sets, the prior art adopts automatic clustering, and clusters similar question sentences to form different question sets. In the process of clustering, it is necessary to determine the number and initial position of the cluster centers to reflect the inter-class dissimilarity of the cluster centers. Then the iterative process of cluster...

AI Technical Summary

Problems solved by technology

[0005] In view of this, the embodiment of the present invention provides a data clustering method and a clustering device, which are used to solve the technical problem of poor clustering effect affected by the initial conditions in the existing problem clustering process

Images