Semantically Enriched Line Search in a Humanities Digital Library

摘要

Digital libraries are proliferating fast. Common examples include domain-specific scholarly repositories like those maintained by reputed technical publishers, digital cultural museums like Europeana, and multi-disciplinary metadata hubs like the National Digital Library of India that indexes a wide spectrum of academic resources. These libraries support different discovery mechanisms but keyword-based search and source-browse remain the most direct and popular ones. However, it is widely recognized in the digital information retrieval community that keyword-based search does not work well when the user's information needs are not well-defined. Although semantic search addresses these issues, it requires complex processing, which is sometimes infeasible. So a digital library designer may explore its limited forms. For example, when a user is looking for works on a topic, her search may be modeled as concept search, which can be supported by indexing the documents against their embedded concepts. Concept search is common in scientific, technical and humanities domains. In this paper, we identify a new variant of limited semantic search specifically suited to humanities. Suppose a research scholar wants to search a humanities digital library with the (probably noisy) first line of a poem. The search engine should retrieve not only the documents containing the line but also critiques on the poem that do not necessarily mention the input line. We call it `semantically enriched line search'. We explain it through examples and describe how this new kind of search may be implemented in a digital library.