Talk:Temporal network

I will move this article to Temporal network as the old title is almost no longer in use in books, searched on Google and appeared almost half as many times in papers found through google scholar. Will hold for a week in case there are any objections. Arashbm (talk) 07:54, 19 July 2018 (UTC)

Ambiguity of some sentences for the average reader.

This article needs revision by a writer who can communicate a complex subject in a clear and unambiguous style. Example of ambiguity; using the word 'second' where it's not clear from the context if the unit of time or the order of events is being referred to. Spyglasses (talk) 16:16, 10 May 2021 (UTC)

User:Arashbm source

Since User:Arashbm has not been inclined to take this up on the talk page as suggested, I will be glad to. The following text and references have been removed from the article by User:MrOllie over concerns that it represents a conflict of interest for Arashbm to add citations to works a work for which he is a contributing author:

Additionally, the event graph representation is used to study spreading processes and existence of time-respecting paths on temporal networks.<ref>Mellor, A. (2018). The temporal event graph. Journal of Complex Networks, 6(4), 639-659.</ref><ref>Saramäki, J., Kivelä, M., & Karsai, M. (2019). Weighted temporal event graphs. In Temporal Network Theory (pp. 107-128). Springer, Cham.</ref> Event graph is a higher-order representation of the temporal network as a [[directed acyclic graph]] where temporal events are represented by nodes and two nodes are connected if they are adjacent, meaning that the two events can form consecutive steps of a time-respecting path. For temporal networks with undirected, instantaneous events, two events are adjacent if they share at least one node and one happens after the other. The event graph representation contains a superposition of all time-respecting paths of the temporal network and can be used to estimate mean and maximum size of an spreading process (e.g. maximum size of an epidemic) from all possible starting nodes and times with a time complexity of <math>O(|E_g| Log |E_g| + |A_g|)</math> as opposed to <math>O(|E_g| |A_g|)</math> of the naive approach, where <math>|E_g|</math> is the number of events and <math>|A_g|</math> is the number of adjacency relationships between all events.<ref>Badie-Modiri, A., Karsai, M., & Kivelä, M. (2020). Efficient limited-time reachability estimation in temporal networks. Physical Review E, 101(5), 052303.</ref>

Temporal Network Theory is a Springer publication, which I would generally take as indicating quality control, and our article on Physical Review E suggests that it is a reliable source. I don't know enough about the field to know whether this content is relevant or useful to the sort of person who would be looking it up here, but I see nothing suggesting that it is misplaced, nor any indication that the editor in question is running amuck adding self-aggrandizing irrelevant content to articles generally. BD2412 T 03:58, 16 June 2021 (UTC)

@David Eppstein: - pinging David Eppstein, on the theory that he, if anyone, can parse the relevance of the proposed text. BD2412 T 04:00, 16 June 2021 (UTC)

Just for clarification, the actual conflict paper is just "Efficient limited-time...", the Physical Review E article. I was not directly involved in the other publications, although my current and last supervisors and some of my current and previous co-workers were. Arashbm (talk) 04:13, 16 June 2021 (UTC)

Corrected. BD2412 T 04:21, 16 June 2021 (UTC)

We already have plenty of sourced material on spreading processes and time-respecting paths on temporal networks. See e.g. the Holme and Saramäki 2012 reference which for some reason we have two copies of. So if the new material is on any added concept, it's on this "event graph", but that appears to be nothing different from a time-expanded network, for which Google scholar has thousands of references stretching back to 1969. So I don't see why we need to use new terminology and brand-new and barely-cited references for this concept. —David Eppstein (talk) 05:43, 16 June 2021 (UTC)

@David Eppstein:, I respectfully disagree that time-expanded networks and event graphs are not different. They are completely separate representations with different properties, where e.g. different things are represented as nodes and nodes are connected under different conditions. I agree that in some cases they can be used to measure the same things, but with that logic, temporal network is also a neologism for dynamic networks or multilayer networks with time slices or just using a bunch of static networks. All the different "good" ways of representing the same phenomena are by necessity equivalent, the difference is the ease to measure interesting characteristic quantity, which is explained in the references and the paragraph. But in any case the conversation is not around the term event graph, but around covering different representations of temporal networks in a section about representations of temporal networks. Even if you are not convinced that event graphs and time-expanded network are distinct way of representing a temporal network, at least of them should be added here. Arashbm (talk) 10:51, 16 June 2021 (UTC)

Given the above, I am comfortable with the determination that the removed content, irrespective of asserted COI, is not beneficial to include in the article. BD2412 T 05:53, 16 June 2021 (UTC)