# networks, social-networks and graph-theory tags

It looks to me like these three tags have a lot of overlap between them - so much that maybe we should to make them synonyms, remove one (or two?) of them or something.

The tag descriptions seem almost interchangeable to me:

• Graphs are abstract representations of objects and their mutual relations, where the objects are 'nodes' and the connections among them are 'edges'.

• Refers to network theory as part of the graph theory. For questions about neural networks, use our [neural-networks] tag.

• Social network data consists of a collection of "nodes" (which can be any sort of entity - e.g. people, corporations) and "links" (which can be any sort of relationship - e.g. friend, sharing a board member).

Out of the 10 newest questions in graph-theory, 5 include the networks tag, and I'd argue that most others could include it too - for example, this one is about doing machine learning (specifically feature extraction) where the data has a graph structure (could just as well be called a network structuree or a social network).

I don't have a concrete suggestion yet - My first though would be to collapse all three into one tag, unless mathematical questions about Graph Theory (not in the practical sense of statistics or learning on networks data) are relevant here (which... I kinda think it isn't?).

• I would suggest thinking about how to apply the tags consistently rather than merging them, because they make potentially useful distinctions: not all questions related to graph theory are about networks or social networks; not all networks are social networks. – whuber Nov 6 '19 at 13:41
• @whuber, that's reasonable, but if those distinctions are too subtle for the typical OP to understand intuitively, the tags will inevitably lead to threads that are less well organized. – gung - Reinstate Monica Nov 6 '19 at 15:04
• @whuber True, not all questions related to graph theory are about networks; but are "pure math" graph theory questions relevant to this site? – Itamar Mushkin Nov 6 '19 at 15:09
• Graph theory, including the parts concerning random graphs, and probabilistic graphs, have many applications that have nothing to do with social networks, including in microbiology, proteomics, genetics, evolutionary biology, linguistics, transit and transportation studies, etc. – Alexis Nov 6 '19 at 18:49
• I would agree that most pure math questions about graph theory are not on topic here. I did answer one about definitions recently (stats.stackexchange.com/a/431261/919), believing that could be useful background for statisticians interested in the subject. The lack of attention (the thread has only 28 views after three weeks) suggests it's not a popular topic here. – whuber Nov 6 '19 at 19:04
• @Alexis - you're absolutely right! But in this contest, the term "social networks" is also used to describe those other applications. From en.wikipedia.org/wiki/Social_network_analysis: "Social network analysis has emerged as a key technique in modern sociology. It has also gained a significant following in anthropology, biology, demography, communication studies, economics, geography, history, information science, organizational studies, political science, public health, social psychology, development studies, sociolinguistics, and computer science." – Itamar Mushkin Nov 7 '19 at 7:08
• @ItamarMushkin We disagree about the history of graph theory and its application in other sciences: while I do not argue that social network analysis has spread beyond its origins in sociology, neither the theory nor the breadth of the methods used in other sciences were developed there. As an example, the analysis of the community matrix – a graph representation of complex causal systems – in population biology and ecology entails several graph analytic approaches, some of which are statistical, and none of which developed from social network theory. – Alexis Nov 7 '19 at 18:51
• I don't think we disagree on the history of the field; at least, I don't disagree with anything in your comment (should this discussion be in an answer? Sorry, new to stats.meta). However, I don't think the history should determine tag partition, but rather current use - especially since the current tag description reflects current use. Let me ask it this way - if "social networks" refers only literally to social networks, why should it have a tag in stats.se? – Itamar Mushkin Nov 9 '19 at 5:28