Modify default network parameters

[Miquel Angel Plaza]

Dear Mr. Chen,
I would like to know your opinion about modifying default network parameters.
I run CiteSpace on a Mac and with version 5.8.R3.
I have tried to obtain a network of co-citations from a set of WoS records. But with default parameters (LRF=3; LBY=5; k=25) no network comes out.
I have varied the k (15,20,25 & 30) and also LBY (8, 10, 15)... but the networks that come out are invalid.

I know that one of the causes of not obtaining a valid network may be that the subject studied is not yet very developed and there are still no relationships to establish a network.

But, finally I tried with this parameters:
LRF=-1
LBY=-1
k=25
And prunning pathfinder and burstness gamma=0.8

And then I obtained what can be a valid network.

My question is if it is a good idea, a serious idea, to analyze a network of this type in which the default parameters of CiteSpace have been varied a lot.

Thanks for your help, if possible
Miquel

[Chaomei Chen]

It sounds the sample size is probably too small for the default setting and that would explain the use of -1s to include everything. This process is a sampling process in nature. The focus is on the most significant portion of your data. Does it answer your question?

[Miquel Angel Plaza]

Many thanks. Yes, I supposed that sample size was too small, but I have done several proofs with CiteSpace default settings and with different sets... one with 583 wos records and another with 1135 wos records... and in both cases I could not obtain useful networks. Not enough clusters, some of them too small and some with Silhouette=1, and no betweenness centrality.
Could be another reason that this field of study is still not well developed and papers are not well related among them?

And another question. Then is is correct to use -1 settings if the network obtained is quite good?

Thanks again for your time.

Miquel

[Chaomei Chen]

Yes, using -1 is certainly acceptable.
If the dataset is small, you may consider expanding the dataset. For example, see below: * C Chen, M Song (2019) Visualizing a Field of Research: A Methodology of Systematic Scientometric Reviews. PLoS One 14 (10), e0223994 * C Chen (2017) Cascading Citation Expansion. Journal of Information Science Theory and Practice 6 (2), 6-23.

[Miquel Angel Plaza]

O.K. Many, many thanks.

Miquel

[Miquel Angel Plaza]

Dear Mr. Chen, I have read with interest both articles you recommended about cascading citation expansion. Both articles shows examples with Dimensions database and through the Dimensions API if I have understood correctly.
And I had read also previously at your handbook "How to use CiteSapce 5.7.R1", p.211, that Cascading citation expansion was done through Dimensions API.

But I have no access to the full database of Dimensions, only free public one.
Is it possible to run cascading citation expansion in WOS or SCOPUS databases?
Do you have some example with WOS or SOCPUS databases?
My computer is a personal one (Mac)... is it possible to run a cascading citation expansion with a personal computer or is it necessary to have a more powerful computer?

If I can not run a cascading citation expansion because I have no access to Dimensions, is it another possible option what you explained at your handbook "How to use CiteSapce 5.7.R1", p. 73? There you explained that in WOS database it is possible to do a Citation Expansion through the Creation Citation Report link but that this would be a less relevant set of records.
Is this another way to improve the original set of WOS records as a way to obtain a better network? Is this a good option?

Thanks
Miquel