FACT-Graph: Trend Visualization by Frequency and Co-occurrence

摘要

In order to visualize keyword trends embedded in newspaper articles, this paper proposes the FACT-Graph (Frequency And Co-occurrence-based Trend Graph). First, we introduce a trend analysis method that works by using keyword classes. We identify four classes of keywords by term frequency (TF) values and document frequency (DF) values in an analytical period, and then some keywords are classified into different classes by period. We pay attention to class transitions between periods and use them as a clue for trend analysis. Next, we apply a method of identifying relationships between multiple words by their co-occurrence and their transitions in order to resolve problems that have occurred in prior class transition analysis. Finally, we output a FACT-Graph by extending the traditional simple co-occurrence graph, which visualizes trend analysis, and simultaneously examine keyword class and keyword co-occurrence relationships. The FACT-Graph is based on four classes of keywords, keyword co-occurrences, and their transitions between time periods. While each class is characterized by the shapes of nodes and keyword co-occurrence relationships are represented by the types of links, the trend transition patterns are colored. Applying the proposed FACT-Graph to a data set of 220,000 newspaper articles, this paper gives some example results and validates the effectiveness of visualizing keyword trends embedded in volumes of text.