Help

Introduction

Silk Road was the focus of attention in the early 20th century. Several large-scale Silk Road expeditions, led by Aurel Stein, Albert Grunwedel, Sven Hedin, and others, performed archaeological, geographical, and cultural studies. As the expedition's achievement, they published volumetric expedition reports that are still essential academic references to research in this area. Recently, Chinese cultural heritage organizations started a large-scale field survey to create a comprehensive database of ruins on the Silk Road with little reference to those expedition reports. The former is valuable because they are the record of the past, which may be destroyed or lost in 100 years. At the same time, the latter is also necessary because they contain up-to-date information about the ruins with more accurate and visual data.

In the Digital Silk Road Project, we have been working on integrating knowledge about Silk Road ruins fragmented in many reports. As the latest development of this project, this paper proposes a system called Silk Road Ruin Database to realize the aggregation of evidence from books and field surveys to create linked ruin data with the help of open source tools. At the core of the database sits a popular web publishing platform Omeka S, with newly developed modules optimized for Silk Road ruins. We also extensively use IIIF (International Image Interoperability Framework) and IIIF Curation Platform to share image data across many providers. The database has six card types, namely place card, place matching card, image card, image matching card, reference card, and ruin card, to aggregate evidence about ruins to create the knowledge graph of ruins or linked ruin data.

Evidence Aggregation

A scenario of evidence aggregation is as follows. Suppose we have data about two ruins from different books or field surveys. We first apply name collation if two ruins have similar names. However, due to the multilingual nature of the Silk Road, a ruin could be named in different languages with little similarity in spelling. Then we turn to place collation to check the proximity of the location of the two ruins. However, due to the error of old maps in expedition reports, the location of ruins may not be accurate enough to decide that the two ruins are the same. Finally, we try image collation to check the similarity of images. If photographs of two ruins look similar, they serve as evidence for the identity of the ruins, even with different names and classifications. This scenario illustrates that the purpose of evidence aggregation is to combine multiple media to increase the reliability of the hypothesis for scholarly research.

Place Collation

We first create the gazetteer of the Silk Road by extracting place names from expedition reports, with the place ID and location coordinates on the old maps in the reports, called Gazetteers of the Silk Road. The gazetteer of expedition reports is then imported into Omeka S using the CSV import module with the automatic creation of a "place card" in the database. The total number of gazetteer entries is 24,139.

Then we start place collation to find the same ruins in different place cards. For this task, we developed a tool called mappinning (map-pinning) to remove the error in the old map. This tool is different from standard georeferencing by allowing interactive single-point georeferencing without distortion of the old map. Mappinning sets the current map (Google Maps) as the base map and an old map as the overlay map. If the overlay map is not pinned, only the overlay map can move; if it is pinned, both maps move together. The error of the old map can be removed by pinning two maps at the corresponding point. After correcting the location, we create a "place matching card" to associate multiple place cards with similar locations. The place matching card may have a unified name, such as concatenating names from collated place cards.

Image Collation

We also take advantage of image data for another type of evidence. The first source of image data is digitized expedition reports from the Digital Archive of Toyo Bunko Rare books. The second source of image data is photographs taken during the field survey on the Silk Road, called Silk Road in Photographs. The former contains old photographs taken about 100 years ago, while the latter contains recent photographs taken during a field survey. We import image data into Omeka S by the following steps. First, we use IIIF Curation Viewer to curate images from IIIF image providers and add metadata to each curated image with a place ID. Then we organize curated images by metadata using Canvas Indexer. Finally, we import the Canvas Indexer data into Omeka S using our newly developed module to automatically create an "image card" in the database.

Then we start image collation to find the same objects in different image cards. We developed vdiff.js (JavaScript-based image collation tool), to compare photographs taken from similar locations and directions. It has two modes, the before-after mode and the transparent mode, to compare an old and a recent photograph. This tool differs from a typical before-after image comparison tool by allowing the real-time projective transformation of images using feature point image matching or manual image transformation. After confirming that the two images are taking the same object, we create an "image matching card" with projective transformation parameters to show the best match of the two images.

Linked Ruin Data

Based on the many cards we have aggregated as evidence, the final step is to conceptualize a unified ruin entity to create a "ruin card," which will be an entry point to the database. Linked ruin data is also a micro-publication platform to demonstrate the collation process toward the conceptualization of a ruin. At present, the number of ruin cards is 49, and we plan to extend this database through international collaboration with more IIIF image providers of the Silk Road, such as European museums, who own a rich collection of digitized historical materials of Silk Road expeditions.

References

  1. 西村 陽子, 北本 朝展, "カード単位の照合エビデンスを共有するシルクロード考古遺跡情報の統合データベース", 人文科学とコンピュータシンポジウム じんもんこん2021論文集, pp. 146-153, 2021年12月
  2. 西村 陽子, 富 艾莉, 北本 朝展, "重識絲綢之路上已発掘古代建築的新方法", 高昌遺珍：古代絲綢之路上的木構建築尋踪 56-69 2021年8月
  3. 西村 陽子, 北本 朝展, "黄文弼地図与欧洲探険隊地図 ——兼及黄文弼所蔵地図調査報告", 西域文史 14 19-47 2020年6月
  4. 西村 陽子, 北本 朝展, 張 勇, "木頭溝的摩尼敎=仏教寺院：絲綢之道遺址数拠庫的建立与遺址核対的深化", 馬可・波羅与10-14世紀的絲綢之道, pp. 172-189, 2019年6月
  5. Yoko NISHIMURA, Erika Forte, Asanobu KITAMOTO, "A new method for re-identifying ancient excavated structures on the Silk Road: the case of Kocho", The Ruins of Kocho: Traces of Wooden Architecture on the Ancient Silk Road, pp. 59-68, 2016年12月
  6. 西村陽子, 北本朝展, "絲綢之路遺址之重新定位与遺址数拠庫之建立", 陝西師範大学学報（哲学社会科学版） 45(2) 76-86 2016年3月
  7. 西村 陽子, 北本 朝展, "地図史料批判に基づくシルクロード都市遺跡・高昌故城の遺構同定", 人文科学とコンピュータシンポジウム じんもんこん2014論文集, pp. 43-50, 2014年12月
  8. 西村 陽子, Erika Forte, 北本 朝展, 張 勇, "古代城市遺址高昌的遺構比定：基于地図史料批判的絲綢之路探険隊考察報告整合", 西域文史, Vol. 9, pp. 153-197, 2014年12月
  9. 西村 陽子, 北本 朝展, "高昌故城調査の統合による探検隊調査遺構の同定―地図史料批判に基づく都市遺跡・高昌の復原―", 『高田時雄教授退職記念学術論文集』 日英文分冊 181-196 2014年6月
  10. 西村 陽子, 北本 朝展, "スタイン地図とGoogle Earthを用いた名寄せと場寄せに基づくシルクロード探検隊調査遺跡の解明", 人文科学とコンピュータシンポジウム じんもんこん2010論文集, pp. 255-262, 2010年12月
  11. 西村 陽子, 北本 朝展, "和田古代遺址的重新定位--斯坦因地圖與衛星圖像的勘定與解讀", 唐研究, Vol. 16, pp. 154-204, 2010年12月
  12. 西村 陽子, 北本 朝展, "スタイン地図と衛星画像を用いたタリム盆地の遺跡同定手法と探検隊考古調査地の解明", 敦煌写本研究年報, Vol. 4, pp. 209-245, 2010年3月

News

2022-11-17
DATABASE OF SILK ROAD RUINS (in English) was released.
2022-10-31
The system and data of Silk Road Ruin Database was updated. Mappinning was integrated into the ruin database, and Gazetteers of the Silk Road was extensively reviewed to have 24,139 placenames.
2022-01-28
Integrated Database of Silk Road Ruins for Sharing the Evidence of Ruin Re-Identification has received the Best Paper Award from IPSJ SIG Computers and the Humanities Symposium 2021.
2021-12-10
New Silk Road Ruin Database was released.
Prev