The relationship between Judaism and Christianity during late antiquity has long been debated, often framed by the simplistic metaphor of a 'parting of the ways.' This research challenges that notion, arguing for a more nuanced and complex interplay between the two religions. Traditional philological methods, while valuable, have limitations in capturing the overall scope and dynamics of these interactions. This study introduces network analysis, a computational approach, to explore the connections between Jewish and Christian communities as reflected in their literary output. Network analysis offers a powerful visualization tool, enabling the examination of complex relationships, identifying key influential texts (nodes), and tracing the flow of information (edges) between traditions. The authors specifically focus on literary interactions, assuming that textual connections between religious communities can reflect actual historical contacts. The research centers on the Babylonian Talmud as the primary Jewish source, complemented by parallel texts from Palestine, and a range of Christian writings from the first to sixth centuries CE. The study aims to demonstrate the potential of network analysis in uncovering patterns and insights that might remain hidden using traditional methods, ultimately providing a more complete understanding of the Jewish-Christian relationship during this formative period.

Prior scholarship on Jewish-Christian relations has often been divided. A large segment disregarded the connection between rabbinic literature and Christian sources, viewing them as distinct entities. Other scholars explored the relationship, focusing primarily on rabbinic works from the land of Israel under Roman rule. Recent research has begun to delve into the Babylonian Talmud in the context of Christian sources, discovering various types of literary interactions, from polemics to analogies and shared theological concepts. However, these studies predominantly employed philological methods, based on manual comparison of textual passages, making large-scale quantitative analyses difficult. Existing network analysis applications within religious studies have mostly focused on social networks (individual interactions) or semantic networks (word interactions), leaving a gap in the study of literary interactions between religious communities. This study bridges this gap by adapting network analysis to explore literary parallels between Jewish and Christian texts.

The research employs a qualitative approach grounded in philological analysis. The authors created inter-religious Christian-Jewish networks. In these networks, nodes represent literary traditions found in either rabbinic or Christian sources, and edges signify a shared tradition or literary interaction between the two. The edges are weighted according to the level of certainty regarding the connection, categorized into three levels: high, medium, and low certainty. Directionality of the edges indicates the source of the information flow (e.g., from a Christian source to a rabbinic one). The authors selected two main sets of Talmudic passages to construct exemplary networks: one from a corpus of stories about heretics interacting with rabbinic figures, and another focused on interactions with contemporaneous Christian monastic traditions. Additional rabbinic parallels from the Palestinian Talmud and Avot deRabbi Natan were included where analogous literary topoi suggested connections. The authors generated weighted directional networks (connectivity matrices) represented graphically in various ways: (1) nodes color-coded by religious origin and geographical location, arranged along a circular time axis; (2) nodes spatially and temporally distributed to visualize the dynamic interactions; and (3) edges color-coded by interaction type (polemic or non-polemic) and weighted by certainty. For one network, Google's page ranking algorithm was used to rank node importance. The data underlying the small-scale networks are presented within the networks themselves, given the limited size of the sample. Detailed explanations of the connections and their weightings are provided in supplementary materials, building upon previous research. The authors acknowledge the inherent limitations in establishing certainty regarding specific connections and dating the interactions, but argue that the network approach offers a novel way to visualize and analyze existing knowledge.

The analysis yielded two primary networks: the Hullin network and the Monastic network. The Hullin network, based on a story in the Babylonian Talmud tractate Hullin 87a, illustrates connections between this text and various Christian sources from different periods and regions. Visual representation reveals the central role of the Talmudic text as a hub, integrating several Christian traditions. A circular temporal representation reveals that the engagement with earlier Christian sources (e.g., New Testament) tends to be more polemical, while engagement with later sources is largely non-polemical. Spatial distribution shows that earlier references are primarily to Eastern Christian traditions, with later references encompassing both East and West. The Monastic network examines literary analogies between Christian monastic texts (such as the Apophthegmata Patrum) and rabbinic traditions. This network reveals the prominent role of the Apophthegmata Patrum, with multiple rabbinic passages showing familiarity with these monastic traditions. The Google page rank algorithm is employed to quantify the relative importance of different nodes. The network also showcases the interconnectedness between rabbinic sources from different geographical locations (Palestine and Babylonia), highlighting the potential of Christian traditions to illuminate intra-religious interactions within Jewish communities. A comparison of the edge-weight distributions between the two networks revealed differences in the strength and distribution of connections, reflecting differences in the certainty of scholarly knowledge and the types of interactions.

The network analysis offers a valuable visual representation of the complex and multifaceted relationship between Judaism and Christianity in late antiquity, moving beyond simplistic narratives of separation. The visualizations reveal intricate patterns of interaction, highlighting key influential texts and the evolution of engagement over time and across geographical regions. The findings challenge the traditional 'parting of the ways' model by demonstrating a more continuous and nuanced relationship than previously acknowledged. The shift from polemical to non-polemical engagement, coupled with the temporal-spatial patterns of familiarity with Christian sources, suggests a dynamic and evolving interaction shaped by time, geography, and the development of distinct theological traditions within both religions. The study also emphasizes the limitations of fully automated text-mining algorithms, advocating for a combined human-machine approach to fully leverage the potential of network analysis in this field.

This paper demonstrates the significant potential of network analysis in illuminating the complex interconnections between Judaism and Christianity in late antiquity. The authors' novel approach provides a visual and quantitative framework for analyzing literary interactions, revealing previously unseen patterns and challenging simplistic interpretations of the historical relationship between the two faiths. The study lays the groundwork for future research employing more comprehensive datasets and advanced network analysis techniques to further refine our understanding of this crucial historical period. Future research could focus on expanding the scope of the networks to include bidirectional interactions, conduct further centrality and cluster analyses, and utilize generative models to predict unknown connections.

The current study is based on relatively small-scale networks, limiting the generalizability of findings to the broader Jewish-Christian relationship. The selection of textual parallels relies on existing scholarly work and may not represent a fully comprehensive sample of all possible interactions. The weighting of edges based on subjective certainty introduces some level of researcher bias, although this is mitigated through detailed explanations and reliance on established research. The small scale of the networks also impacted the ability to perform complex statistical analyses on weight distribution. The authors acknowledge the need for future research with larger datasets and advanced computational methods to further explore and refine these findings.