Embedded Files
The following network is also based in the co-occurrence of two keywords within the same document. The graph can be consulted by:
searching a keyword troughout the Search bar in the grey panel;
selecting a whole group with the Group Selector;
clicking directly a keyword within the network;
A complementary Information Pane will be opened on the right side, where statistic network analysis are displayed, given the Modularity Class, Eccentricity, Closeness Centrality, Betweenness Centrality, Harmonic Closeness Centrality, and Degree; as well as the group members, in the case of selecting clusters.
Picture 8. Network analysis of the most frequent keywords of the sample based on their coocurrence within the documents. The network was obtained through bibliometrix package and deployed with Gephi (Bastian & Ramos Ibañez, 2017) and sigmaJs (Jacomy, 2017).
The central node is withouth doubt digital humanities, which connects all the nodes. Its best connection is visualization, following by digital libraries. The first one represents better the more technical approaches including machine learning; meanwhile the second is better connected with cultural heritage. However, all of them belong to the PURPLE CLUSTER (55 members), which is, with great difference, the larger cluster. There are just a few small cluster in adittion to the main one:
The ORANGE CLUSTER (4 members): with technical issues on text processing, topic modeling, text analysis, literature and ocr.
The GREEN CLUSTER (7 members): more related with humanities, interdisciplinarity, epistemology, culture or sustainability.
The RED CLUSTER (6 members): about social relations and knowledge communication, with keywords such as social media, open science and citizen science.
The YELLOW CLUSTER (6 members): the technical approaches of network analysis, tei and stylometry.
The BLUE CLUSTER (2 members): virtual and augmented reality.
Centrality Measures
Centrality Measures
In addition to the network visualization, there are internal measures that can provide interesting information about the network behaviour. Below, a table containing some of the main measures that characterizes the network relations. In this case, since the previos visualization was made representing the weighted degree of the nodes, the rows were ordered following the betweenness centrality. It expresses the capacity of the nodes to reach faster, or in a shorter way, more nodes. So, through betweenness, the most intermedial nodes are detected, those that having or not more amount of connections, they connect better and faster with more.
Table 2. Nodes of the author keyword network on Digital Humanities following the betweenness centrality. The weighted degree expresses the total number of connections that reach a node, including their frequency. Other measures of centrality shown are: closeness, that measures the average distance from an initial node to all other nodes in the network; and eccentricity, that measures the distance from a node to the node furthest away from it in the network.
It is also interesting to observe when the order of betweenness centrality changes with respect to the weighted degree order, which cells are colored following its rank. As it can be seen, keywords such as metadata, history, humanities, or database (or even, below, digital libraries) have high betweenness centrality despite their not so high weighted degree. These four keywords are pointing out to classical subjects (humanities and history) as well as to the manage of data (metadata and database). That are very transversal issues within the data sample, highlighting the concern about positioning humanities proper contents within the digital revolution.
Page updated
Report abuse