Florian Atzenhofer-Baumgartner,Bernhard C. Geiger,Christoph Trattner,Georg Vogeler,Dominik Kowald
Abstract:This extended abstract describes the challenges in implementing recommender systems for digital archives in the humanities, focusing on <a class="link-external link-http" href="http://Monasterium.net" rel="external noopener nofollow">this http URL</a>, a platform for historical legal documents. We discuss three key aspects: (i) the unique characteristics of so-called charters as items for recommendation, (ii) the complex multi-stakeholder environment, and (iii) the distinct information-seeking behavior of scholars in the humanities. By examining these factors, we aim to contribute to the development of more effective and tailored recommender systems for (digital) humanities research.
What problem does this paper attempt to address?
This paper attempts to address the challenges faced in implementing recommender systems (Recommender Systems, RecSys) in digital archives in the humanities field, especially for historical legal documents (i.e., "charters") on the Monasterium.net platform. Specifically, the paper explores the following three main issues:
1. **Historical Documents (Charters) as Unique Items for Recommender Systems**:
- Historical documents are complex and multifaceted, containing different levels of integrity and authenticity. They are not only legal documents but also have the value of cultural heritage.
- These documents exist in the form of digital scans, academic versions, and semi - structured data, and are represented using specialized encoding schemes (such as Charter Encoding Initiative, CEI). These representations include diverse metadata, such as material properties, production details, verification methods, and academic annotations.
- The high - dimensionality and sparsity of metadata pose significant challenges to traditional recommendation algorithms.
- Since the number of historical documents is limited and spans different historical periods, the recommender system needs to consider time - relatedness.
2. **Digital Archives in a Multi - Stakeholder Environment**:
- Implementing a recommender system must take into account a complex ecosystem of stakeholders, each with different values and goals. Key stakeholders include researchers (such as historians and other humanities scholars), content creators (such as archivists and editors), as well as platform owners and funding agencies.
- Balancing these diverse and sometimes conflicting interests is a major challenge. For example, giving priority to recommending popular or well - documented historical documents may meet the needs of ordinary users, but will limit the exposure of less - known but crucial documents for specific research.
3. **Information Retrieval Behavior in Humanities Research**:
- The information retrieval behavior of humanities scholars is significantly different from that of users in commercial recommender systems. For example, historians usually conduct exploratory searches, aiming to discover unexpected connections and generate new research questions.
- The concept of relevance in humanities research is multifaceted and context - dependent, which makes it difficult to apply traditional recommender system relevance measures.
- Due to the long - term nature of humanities research, the interaction patterns between users and the system are also different. The value of recommendations may only become apparent after a long - term study.
By addressing these issues, the paper aims to provide directions for developing more effective and customized recommender systems, thereby enhancing the practicality of digital archives and promoting new discoveries and in - depth understanding in historical research.