Contextual Document Similarity for Content-based Literature Recommender Systems

To cope with the ever-growing information overload, an increasing number of\ndigital libraries employ content-based recommender systems. These systems\ntraditionally recommend related documents with the help of similarity measures.\nHowever, current document similarity measures simply distinguish between\nsimilar and dissimilar documents. This simplification is especially crucial for\nextensive documents, which cover various facets of a topic and are often found\nin digital libraries. Still, these similarity measures neglect to what facet\nthe similarity relates. Therefore, the context of the similarity remains\nill-defined. In this doctoral thesis, we explore contextual document similarity\nmeasures, i.e., methods that determine document similarity as a triple of two\ndocuments and the context of their similarity. The context is here a further\nspecification of the similarity. For example, in the scientific domain,\nresearch papers can be similar with respect to their background, methodology,\nor findings. The measurement of similarity in regards to one or more given\ncontexts will enhance recommender systems. Namely, users will be able to\nexplore document collections by formulating queries in terms of documents and\ntheir contextual similarities. Thus, our research objective is the development\nand evaluation of a recommender system based on contextual similarity. The\nunderlying techniques will apply established similarity measures and as well as\nneural approaches while utilizing semantic features obtained from links between\ndocuments and their text.\n