What's New in 6.3.R3 Advanced

Updated 6/27/2024

Timeline View

The Timeline View has been updated. Node labels are getting a bit closer to their nodes. You may choose a background color for your node labels. Here is a fisheye view in the Timeline View so that nodes in the recent years are not too crowded together.

Google Earth KML

CiteSpace uses https://geocode.maps.co for geocoding purposes. The Geocode.maps now requires an API key. You need to obtain your own API key from them to use this function in CiteSpace. Set an environment variable GeocodeAPI with your own API key.

There are a few plans. Currently, CiteSpace assumes you have the free API, which limits to 1 request per second and a total of 100,00 per month.

Scopus

Author affiliations are originally formatted differently in Scopus from the Web of Science. In Scopus, the institution name is often buried in the middle of a long string of a department, a college, its city, and country. If you need to generate a network of collaborating institutions, you may find some lower-level units such as departments appear as nodes instead of institutions. In 6.3.R3, the Scopus conversion function is revised to take this issue into account, at this point, most of time. In future releases, the user should be able to make additional changes.

Author Labels

Some users asked how to show more node labels in a co-author network. The key is the threshold in the Control Panel. Before clustering, the default mode is to show node labels based on their frequencies. The notion of frequency is context sensitive. In a co-author network, the frequency is the number of qualified publications associated with an author node. In a network of cited references, the frequency is the time of times the reference has been cited. The default mode is By Citation, which may make things a bit more puzzling for beginners. If you want to see more labels, simply lower the threshold value.

The default mode will change to By Cluster if you generate clusters. Once clusters are identified, node labels are shown By Cluster, which means the same number of top-cited or top-whatever nodes will have their labels shown. This option will present information evenly across clusters.

Dual Map, Heatmap, Landscape, Timezone, Cluster View

None or minor changes are made for this group of functions.

Ollama

Last, for the adventure seekers, as mentioned in another blog, a new option is added for using Ollama in additional to the default GPT from OpenAI. You will need to have Ollama setup on your localhost:11434 or have it setup on a remote server and provide the endpoint when asked. You will also need to provide the name of the model you would like to use, for example, llama3. As far as CiteSpace is concerned, that is all you need to tell CiteSpace. Beyond that, you will be essentially on your own to figure out the Ollama side of the setup.

A new patch of CiteSpace was released on 6/27/2024. Here are some examples with llama3 and gemma:2b.

Cluster Labels: llma3 vs gemma:2b

#0: Trauma Response vs Post-9/11 Trauma in Manhattan

#1: Biological Terrorism Threats vs Bioterrorist Attacks on Food

#2: Chemical Attack Consequences vs Emergency Medical Responses to Chemical Attacks

#3: Disease Prevention vs Bioterrorism and Emerging Infectious Diseases

#4: Bomb Blast Injuries vs Blast-Related Injuries

llma 3 (4 GB):

gemma:2b (1.7 GB)

Cluster Summaries: llama3 vs gemma:2b

llama3:

gemma:2b:

Currently, I only tested llama3 and gemma:2d. The responses from gemma:2b are slightly more specific than the output from llama3. Overall, they both appear to be meaningful.