Dynamic burstiness of word-occurrence and network modularity in textbook systems

• Published in: Physica A Vol. 487; pp. 103 - 110
• Main Authors: Cui, Xue-Mei, Yoon, Chang No, Youn, Hyejin, Lee, Sang Hoon, Jung, Jean S., Han, Seung Kee
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.12.2017
• Subjects: Learning dynamics; Network community structure; Physics knowledge structure; Physics terminology network; Temporal analysis; Temporal analysis; Physics knowledge structure; Physics terminology network; Learning dynamics; Network community structure
• ISSN: 0378-4371; 1873-2119
• DOI: 10.1016/j.physa.2017.06.002

We show that the dynamic burstiness of word occurrence in textbook systems is attributed to the modularity of the word association networks. At first, a measure of dynamic burstiness is introduced to quantify burstiness of word occurrence in a textbook. The advantage of this measure is that the dynamic burstiness is decomposable into two contributions: one coming from the inter-event variance and the other from the memory effects. Comparing network structures of physics textbook systems with those of surrogate random textbooks without the memory or variance effects are absent, we show that the network modularity increases systematically with the dynamic burstiness. The intra-connectivity of individual word representing the strength of a tie with which a node is bound to a module accordingly increases with the dynamic burstiness, suggesting individual words with high burstiness are strongly bound to one module. Based on the frequency and dynamic burstiness, physics terminology is classified into four categories: fundamental words, topical words, special words, and common words. In addition, we test the correlation between the dynamic burstiness of word occurrence and network modularity using a two-state model of burst generation.