An Improved K-means Service Clustering Method Based on Topic Modeling

Abstract

An improved K-means service clustering method around topic modeling, including the following steps: the first step, preprocessing all mashup service data that needs feature representation; the second step, based on the preprocessed mashup service data , to extract the functional nouns; the third step, for the functional noun set FS of each mashup service, use the topic model to represent the mashup feature vector; the fourth step, for all the mashup feature vectors participating in the clustering, carry out the density information Calculation; the fifth step, based on the density information calculated in the fifth step, select the candidate points of the cluster centers from all the Mashup feature vectors; the sixth step, further screen out the candidate points of the cluster centers obtained in the fifth step The most suitable K initial cluster centers are used for K-means clustering. The present invention improves the final effect of Mahsup service clustering.

Description

technical field [0001] The invention relates to the field of Mashup service data clustering in the Web environment, in particular to an improved K-means clustering method around topic modeling. Background technique [0002] As one of the core technologies of the Web 2.0 era, Mashup technology can realize the integration of heterogeneous resources by combining WebAPI services with different functions. Once this convenient and efficient development technology came out, it has been favored by the majority of software developers, and many organizations have also released their own developed Mashup services and data resources to the Internet for users to call. However, with the continuous growth of mashup service resources on the Internet, how to help users quickly locate mashup services that meet their own needs has become an urgent problem to be solved. In addition, because most of the current Mashup services lack normative WSDL documents and related service attribute descript...

AI Technical Summary

Problems solved by technology

Although the idea of ​​the DPC algorithm is concise and efficient, there are still some problems in practical applications: (1) the clustering effect is very dependent on the selection of the cut-off distance; (2) when the amount of data is large, it may not be easy to pick out Appropriate cluster center point

Images