Big Data & Society 7(2), DOI https://doi.org/10.1177/2053951720968865. First published: Nov 11th, 2020.
Understanding the heritage market, which includes sales of portable objects from the ancient to more recent past, has been a challenge for heritage experts. In part, this is because the market is not only dynamic but has also included illegal sales that are difficult to track. This study attempts to understand the heritage market by using a potentially important proxy, specifically eBay’s site for selling heritage on the international market. By looking at eBay, the relatively large volume of sales can be monitored to determine patterns of sales that include what cultures are sold, the types of objects sold, and the materials that they are made from. Additionally, it is possible to determine where items are sold from, including what country are selling given objects. While sales data provide us with potential insights that may inform on the wider market, what is also useful is we can use a site like eBay to see how clearly cultural value may potentially drive sales. Cultural value is defined as the appreciation or beliefs one develops about given cultures, including ancient ones, from media, educational, and other resources. Effectively cultural value is driven by our experiences in our modern lives that make us value given cultures, including ones from the ancient past, more than others. The link between cultural value and monetary value is potentially a strong link, as the results of this work show.
Data on eBay sales are partially unstructured. This requires an approach that not only can obtain relevant data but can apply machine learning methods to determine relevant terms for describing information about cultures, types of items, and materials for given objects sold. Named entity recognition (NER) has been a category of methods that attempts to meaningfully categorise terms that could be summarised for information, including quantitative methods that summarise wider term patterns. Conditional random fields (CRFs) are one way to apply NER, where word patterns can be recreated as an undirected graph and the probabilities of terms associated with other terms can help to categorize a given term. For instance, Iron Age can be categorised as a period, based on the association of Age with Iron. However, iron in the terms iron sword can be categorised as a metal given its association with sword. These term associations help to categorised such unstructured descriptive terms that are provided as part of the sale data on eBay. Additionally, using term dictionaries can assist with the fact that sometimes the lack of training data may make categorisation more difficult. Overall, the approach applied in this work allows us to relatively accurately categorise the cultures, types of objects, and the materials objects are made from. Structured data within eBay also allow us to know the countries where sales happen, the price items sold at, and even the sellers (which was anonymised in the article but tracked for statistical purposes).
For the study period (October 21, 2018-January 3, 2020), countries such as the UK, US, Cyprus, Germany, and Egypt are among leading sellers of antiquities, or heritage objects, on eBay. These countries have shown strong cultural value for the cultures that appear to sell the most on eBay. Cultures such as the Romans, ancient Egypt, Vikings (or Norse/Danes), and the ancient Near East have been a fixture in Western education and media and are the top selling cultures. Additionally, items such as jewellery, statues and figurines, and religious items sold the most; masks and vessels (e.g., such as vases) may not have as high a volume in sales but generally fetch higher prices. Metal, stone, and terracotta are also the most commonly sold materials. On the other hand, ivory, papyrus, and wood obtain some of the higher prices. What is also clear is that sales are often driven by a relatively small number of sellers, with about 40% of sales over part of the study period dominated by 10 sellers. Sales disproportionally concentrate in Western countries, but emerging markets such as Thailand are evident. Many countries that dominate sales in one category of items also dominate sales for other objects (Figure 1). Some countries, however, are also specialists with certain cultures or object types, such as Canada, Latvia, India, and Israel. The key finding in this work is we can see that cultural value has a link with sales on eBay and eBay, at least in part, can act as a proxy for wider antiquity sales given that it seems to demonstrate Western markets dominate sales, similar to what has been noticed anecdotally. Other sites and even social media, in the long term, would need to be monitored for their heritage sales to obtain a fuller idea of the market. This work is a start, with the code and data used provided as part of the outputs.
Figure 1. Maps showing where cultures were sold (a) and total sales for countries (USD).