Transport networks and towns in Roman and early medieval England: An application of PageRank to archaeological questions

Abstract

This paper examines the development of a road network through time to consider 8 its relationship to processes of urbanisation in Roman and early medieval England. 9Using a popular network measure called PageRank, we classify the importance of 10 nodes in the transport network of roads and navigable waterways to assess the 11 relative location of urban places. Applying this measure we show that there is a 12 strong correlation between the status of towns in both Roman and medieval 13 periods and their proximity to transport nodes with high values of PageRank. 14 Comparison between two temporally distinct networks-Early Roman, and that 15 recorded in the Domesday survey of AD 1086-allows for a discussion about the 16 determinants of urban growth at different times. The applicability of PageRank to 17 other forms of network analysis in archaeology are offered in conclusion. (Gross and Sayama, 2009;Lewis, 2009; Wasserman and 26 Faust, 1994) to other fields have yielded interesting results and proven that promising areas 27 of research can emerge (see for e.g. Barabasi and Oltvai, 2004;Barthelemy, 2011; Borgatti et 28 al., 2009;Dunne et al., 2002;Emirbayer and Goodwin, 1994;Glaisyer, 2004; McPherson et al., 29 2001;Proulx et al., 2005;Singh, 2005;Sorenson et al., 2006;Watts and Dodds, 2007). In 30 archaeology, there is a burgeoning literature utilising such approaches (see, for e.g. 31 Broodbank, 2000;Brughmans, et al., 2016;Collar et al., 2015;Hage and Harary, 1991; 32 Knappett, 2011 32 Knappett, , 2013. Notably, these have included attempts to define network properties 33 from distributions of archaeological materials such as physical monuments (e.g. Johansen, et 34 al. 2004), artefacts of known provenance (e.g. Sindbaek, 2007), historically-attested journeys 35 or physical route networks (e.g. Graham 2006;Isaksen, 2008), and other cultural phenomena 36 such as language and legal traditions (e.g. Collar, 2013;Terrell, 2010). 37 38 Relevant to this paper, are works that have applied network analysis to explore the structure 39 of past transport networks. In an early application of this type, Dicks (1972) evaluated the 40 Roman road system of Britain via path ordering. More recently, Orengo and Livarda (2015) 41 have tested the same evidence to demonstrate the relation between the trading activities of 42 towns and the connections of nearby transport links. 43 44 Extending these debates, this paper seeks to demonstrate the applicability of using a popular 45 network measure called PageRank to archaeological questions. An algorithm used by Google 46 Search, PageRank (Brin and Page, 1998) was originally designed as a way of measuring the 47 relative importance of web pages based on the links among them. The intended purpose of 48 such ranking was to filter the web pages and return the most relevant ones in response to a 49 query given to the search engine. However, the algorithm itself is universal and can usefully 50 be applied to other situations where one seeks to identify the importance of entities in a 5...

Understanding the impact of network structure on air travel pattern at different scales

“…similar to the case of other urban transport network [50]. It can be argued that in this PageRank model, a node benefits more from having direct connection with an important node in the network.…”

A data-driven optimization of large-scale dry port location using the hybrid approach of data mining and complex network theory

“…Compared to other centrality indices such as degree centrality which measures the node importance simply by counting number of neighbours connected with the node, many studies of transportation networks favor EC since it can provide more profound insights about the node influence in the network (Brookes and Huynh, 2018;El-adaway et al, 2018;Parajuli and Haynes, 2018).…”

Patent Cooperative Patterns and Development Trends of Chinese Construction Enterprises: A Network Analysis

“…It can be used to find the most influential node in the network (Google, 2001). Various studies had tested different damping factors, but it was assumed that the damping factor would be approximately 0.85 (Brin & Page, 1998;Brookes & Huynh, 2018).…”