The rapid growth of digital resources in modern digital libraries has intensified information overload, particularly in collections dominated by periodical publications. Such materials are inherently heterogeneous, multi-thematic, and often noisy, which complicates document modelling, similarity assessment, and the delivery of personalized content. The purpose of this research is to design and empirically validate a multi-component document similarity measure that supports robust and interpretable personalization in periodical-heavy digital libraries. Existing similarity approaches are largely designed for short or single-topic documents and, therefore, struggle to capture the partial and localized thematic overlap that characterizes periodicals. To address this gap, this study proposes a multi-component document similarity measure explicitly tailored to long, multi-topic periodical content. The proposed measure combines complementary perspectives on document relatedness by jointly accounting for global thematic orientation, localized content overlap, thematic distribution, and factual context within a unified and parameterizable formulation. By treating similarity as a composite phenomenon rather than a single-dimensional score, the approach enables more stable and meaningful identification of related documents in heterogeneous collections. The proposed measure is intended to support similarity-based navigation by helping users locate documents related to a specific document they are currently accessing or reviewing, thereby improving focused exploration within complex digital library collections.