A Method of Text Mining Based on Fuzzy Neighborhood and Application to Twitter Analysis

Abstract

We introduce fuzzy neighborhoods model to analyze the text as a sequence of words. While the conventional model considers only the number of occurrences of the words, fuzzy neighborhoods model considers the sentence structure. We prove that a class of functions representing fuzzy neighborhoods is a kernel function. We perform term clustering with hard c-means using the kernel function. We handle real data of the Twitter, and compare result with different size of the neighborhoods.