VISUALIZATION FOR CHANGES IN RELATIONSHIPS BETWEEN HISTORICAL FIGURES IN CHRONICLES - Information Visualisation: Techniques, Usability and Evaluation - page 147

5.1 Extracting Clusters of Keywords

A network of keywords is extracted from 1560 to 1580 by replacing a
person with a keyword in equation 1 in Section 4. 3 A set of nodes
(keywords) directly connected by edges is extracted as a cluster. This
categorizes 3034 keywords in 70 clusters. Table 8.1 shows examples of
clusters that use the keyword with highest frequency as a cluster label.

Cluster
label

Keywords

Battle Ikko, Shin Buddhism, Follower, Battle, Rebellion,
Crucifixion, Naval battle, Navy, ...
Gift Whale meat, Gift, Persimmon, Wild goose, Garden
lantern, Present, ...
Allowance Loyalty, Allowance, Supply, ...
Grant Incense, Grant, ...
Conferment Commission, Conferment
Succession

Successor, Succession, Head of family

... ...

Table 8.1. Example of keyword clusters.

5.2 Extracting Clusters for each Connection between Historical
Figures

Next, we extract clusters and their frequencies every year related to
each person dependency (edge) mentioned in Section 4.
More specifically, records are collected in which two persons related to
each edge appear at the same time span. We then obtain keywords
appearing in the collected records and obtain clusters and frequencies for
the edge by using the result of keyword clustering described in Section
5.1. Table 8.2 gives an example of extracted clusters and frequencies for
the edge between ODA Nobunaga and TAKEDA Katsuyori 4 in each year.

3. We use Į = į = ȝ =0.8 in this case.
4. TAKEDA Katsuyori was a famous rival lord of Nobunaga