====== To do ====== ===== CociteSeer ===== Tsung Teng Chen, David C Yen: CociteSeer: A system to visualize large cocitation networks. The Electronic Library · August 2010, DOI: 10.1108/02640471011033602 Shrink groups of nodes of degree 2 (small degree) to get an overall structure. ===== Extreme Multilabel Classification ===== * ... and logistic regression for multilabel classification - ?Cheng - [[http://scholar.google.si/scholar_url?url=http://www-old.cs.uni-paderborn.de/fileadmin/Informatik/eim-i-is/PDFs/ml09draft.pdf&hl=en&sa=X&scisig=AAGBfm2L5LISMMRne9mQUHqxl3YTtC84Hg&nossl=1&oi=scholarr&ved=0ahUKEwjyrb_-yNLYAhVBniwKHf0XCqkQgAMIKCgBMAA|PDF]] * … local embeddings for extreme multi-label classification - ?Bhatia - [[http://scholar.google.si/scholar_url?url=http://papers.nips.cc/paper/5969-sparse-local-embeddings-for-extreme-multi-label-classification.pdf&hl=en&sa=X&scisig=AAGBfm11lzK9VA7ROdwjVnFAlDz04ozJVg&nossl=1&oi=scholarr&ved=0ahUKEwjyrb_-yNLYAhVBniwKHf0XCqkQgAMIKSgCMAA|PDF]] * http://manikvarma.org/downloads/XC/XMLRepository.html * http://www.machinedlearnings.com/2015/04/extreme-multi-label-classification.html * https://link.springer.com/chapter/10.1007/978-3-319-64283-3_21 * http://www.wikicfp.com/cfp/servlet/event.showcfp?eventid=71852©ownerid=5960 * http://riemenschneider.hayko.at/vision/dataset/task.php?did=316 * http://www.uni-obuda.hu/users/szakala/SMC%202016%20pendrive/2032_smc2016.pdf * http://proceedings.mlr.press/v70/si17a.html * [[https://www.polibits.gelbukh.com/2016_54/IN-DEDUCTIVE%20and%20DAG-Tree%20Approaches%20for%20Large-Scale%20Extreme%20Multi-label%20Hierarchical%20Text%20Classification.pdf|IN-DEDUCTIVE and DAG-Tree]] * https://www-cs.stanford.edu/~jure/pubs/node2vec-kdd16.pdf * https://sourceforge.net/p/meka/mailman/message/35906727/ * https://www.dagstuhl.de/de/programm/kalender/semhp/?semnr=18291 ===== Network multiplication with threshold ===== December 14, 2017 Skip small values. Introduce a threshold value. At the end correct values of existing links. ===== Leaders approach to clustering with RC ===== December 14, 2017 Dissimilarities are computed between leaders. ===== Corrected fractional bibliographic coupling ===== January 2018 Jaccard(A,B;k) = |A ∩ B| / max(k,|A ∪ B|) Poglej: http://gtna.cs.uni-kl.de/en/publikationen/PDFs/ASystematicApproachToTheOneModeProjectionOfBipartiteNetworks.pdf ===== Intensity of node activity in temporal networks ===== January 2018 u -> v : v(i) = u(i)/(t(i)-t(i-1)) ===== Wasserstein and TQ ===== March 20, 2018 Can the Wasserstein distance be combined with TQ? Interval -> Quantiles ===== Clauset ordering of journals ===== March 21, 2018 Compute JCi = JW * Ci * WJ . What is the condensation structure (of the main weak component)? Apply Clauset ordering to it. JA = JW * WA. What is "the right" normalization of JAJ = JA * AJ ? ===== Symbolic data ===== March 31, 2018, Piran Let S1 = ∑ xi and S2 = ∑ xi2. We define μ = S1/n and σ2 = ∑ (xi - μ)2 / n = S2 / n - μ2 . Instead of μ or ( μ, σ ) it is better to use ( n, μ ) or ( n, μ, σ ) because we can compute also the representative of a union of disjoint sets X and Y, X ∩ Y = ∅ :\\ nX∪Y = nX + nY\\ μX∪Y = (nX⋅μX + nY⋅μY) / nX∪Y \\ σX∪Y2 = S2;X∪Y / nX∪Y - μX∪Y2 \\ where: S2;X∪Y = S2;X + S2;Y and S2;X = nX⋅(σX2 + μX2) https://en.wikipedia.org/wiki/Standard_deviation Ward's clustering method has a criterion function P(**C**) = ∑C∈**C** σC2 . ===== Brain networks ===== Is there a fast criterion to decide that a node is isolated (of degree 0)? ===== Citation analysis ===== April 25, 2018 Extend bibliography from "Exploring the Limits of Complexity: A Survey of Empirical Studies on Graph Visualisation" (review for Information Visualization). August 20, 2018 * Use node weights in citation networks for analysis * Does Kirchoff law hold for node weights? * Standardization of any network: link s to all IN-nodes, link all OUT-nodes to t, add (t,s) arc. The standardized network is strongly connected. It can be periodic. A loop on (t,s) arc? ===== Bibliographic analysis ===== See {{notes:doc:1findr.pdf}}. Analysis of works using important "keywords" from text. ===== Small world ===== Moskva, May 27, 2018 Milgram paper. Inbreeding can lead also some levels back. Clustering coefficient considers only one level inbreeding. ===== Configuration model ===== Moskva, May 29, 2018 Chung-Lu method for large networks. Each row u (or column) has the same partial distribution indeg[v]/m. Select randomly in each row u outdeg[u] nodes following this distribution using an improved tabelaric method - merging a sequence of basic random numbers? If you get duplicates or loops skip them. See [[https://scholar.harvard.edu/joelmiller/random-network-generation|Miller]], [[https://scholar.harvard.edu/files/joelmiller/files/fast_chung_lu_generation.pdf?m=1413848771|paper]], [[http://networkx.lanl.gov/archive/networkx-1.6/reference/generated/networkx.generators.degree_seq.expected_degree_graph.html|code]]. Can be extended to https://mathinsight.org/generating_networks_second_order_motif_frequency ? See regularity measures in Module 2 at http://www.pitt.edu/~kpele/tele2125_spring15.html#lectures ===== Markov chains ===== Moskva, June 2, 2018 Markov chains can be used to summarize sequences of events. For example sequence of letters in text -> transition matrix * generate random sequence determined by a MC * can we use MC to measure the conformance of a sequence to MC ? Would this work: assign to each sequence as its value the geometrical mean of MC transition probabilities? Small values - atypical sequence; 0 value - forbiden sequence. * normalization of the value? multiply it by number of states n ? [[https://stats.stackexchange.com/questions/247675/calculate-the-probability-for-a-sequence-generated-by-a-graph|StackExchange]] ===== Network measures ===== Moskva, June 4, 2018 Zhukov's slides: Lecture 5/slide 8: Harmonic centrality; Lecture 7/slide 12: Regularity measure !? ===== Topogical approach to networks ===== LJ, November 8, 2018 Try to extend the approach from prof. dr. Robin Henderson, Newcastle University: Statistical Topology and the Random Interstellar Medium (8. november 2018 - 11:00, IBMI) to networks. See photos. The results are the same for all monotonically increasing transformed values?! ===== Genealogy of the field ===== LJ, November 11, 2018 How to produce the genealogy (Figure 1) from [[https://www.researchgate.net/publication/328574199_Spatial_Decision_Support_Systems_Three_decades_on/references|Spatial Decision Support Systems: Three decades on]] {{vlado:notes:pics:sdss.png}} ===== Dictionary networks ===== Moscow, December 6, 2018 Use n(Dic)*t(n(Dic)) or d(X,Y) = |X ∩ Y| / min(|X|, |Y|) to define a (dis)similarity and determine the islands. ===== Network spreadsheet ===== Moscow, December 6, 2018 Igraph node/link attributes can be also structured object. Develop computation engine for values and functions on nodes/linka. ===== Function form ===== LJ, Januray 5, 2019 Express the function from the [[https://tech.paulcz.net/blog/future-of-kubernetes-is-virtual-machines/|blog]] / [[https://commons.wikimedia.org/wiki/File:Gartner_Hype_Cycle.svg|picture]]. Lognormal + natural growth (max,min,lim,shrink) ===== Interesting ===== Brian Nguyen: Multivariate Statistical Analysis with R: PCA & Friends making a Hotdog https://bookdown.org/brian_nguyen0305/Multivariate_Statistical_Analysis_with_R/ ===== Temporal groups ===== 14. Jul 2022 A temporal group is formed around a "leader" (generator, attractor). ===== Specificity ===== 28. Mar 2024 # citatitons (w) inside the topic bibliography / total # of citations (w) inside OpenAlex ===== TF-IDF ===== 29. Mar 2024 topic bibliography vs. OpenAlex