Text Cluster Analysis Based on Haikou 12345 Hotline Data

Abstract

Haikou 12345 hotline is a system that processes a large number of help requests or complaints from the citizens in the city. The paper resorts to the concept of cluster-based machine mining approach to facilitate the process. According to the clustering results of the incoming texts in the system, governmental resources can be prioritized to handling the most publically important issues. In other words, the government personnel could have a better focus. Clustering is conducted dynamically in real time such that the decision-making can be performed in an optimal way to use the government resources, e.g., labor. In order for clustering, the system is designed the functions for preprocessing the data, data cleaning, Chinese word segmentation, and document deduplication before learning. Clustering uses a faster K-means algorithm to obtain more complete and efficient clustering results in an effective time.