This dataset consists of point data and line data, and its geographical coverage is limited to regions along the Silk Road.
Point data were extracted from documentary records. By referring to archaeological materials and other research results,3 most of the travel sites were accurately located and correlated to a specific town and site of modern times. To the utmost of the authors’ capability, we paid field visits to a few places in the case when visits were deemed necessary. Once the specific location of the sites were identified, we used Google Earth to locate their current position, by referring to the positioning theory as per “GIS technology and the study of Silk Road restoration over 2000 years”.4 Take Beshbaliq as an example: existing scholarship shows that the place was located in the present-day Pochengzi ruins, Beiting town, Gimosar, Changji, Xinjiang.5 Through field visits (Figure 1) and satellite imagery, we confirmed the specific location of the site (Figure 2). After the places were located, KML data were formed, which were then edited in ArcGIS and exported into 20 groups of point data.
Figure 1
Field investigation of the historical site of Beiting Figure 2
Site location using Google Earth satellite images Based on documentary records, we referred to the Silk Road restoration methods to restore line data, as delineated in “Research on the GIS-aided restoration of Linzhou Passage in the early Northern Song dynasty”.6 Based on the point data collated, routes were drawn on Google Earth by referring to field investigation findings, historical and modern maps, satellite images, and other research results.7,8 Take the Guazhou-to-Shazhou section as an example. Through field investigations (Figure 3), the authors got to know that this section extends along the northern foot of the mountain range, and crosses the desert and oasis. We referred to the route direction as recorded in Mongolia Landscape Atlas (Figure 4), as well as on-site investigation data, based on which a route consisting of several sections was drawn against the background of Google Earth satellite image (Figure 5). The data were edited and categorized into 20 KML files according to the tourists’ itinerary, which were further edited in ArcGIS and exported into 20 groups of line data.
Figure 3
Field study of the Guazhou to Dunhuang section Figure 4
Roads passing through Shazhou as recorded in Mongolia Landscape Atlas Figure 5
Route restoration by reference to Google Earth satellite image and the acquired data After the arrangement of point data and line data, the data files were named after the start time of the itineraries recorded. Shapefiles of point data were designated by the “points” suffix, while those of line data were designated by the “lines” suffix. All the data were grouped into 19 periods, of which two tourists set off in 1260 and were thus marked by "-1" and "-2" in the respective file names as a way to distinguish. The main process of data generation is shown in Figure 6.
Figure 6
Flowchart of dataset generation 
