Semantic Query Expansion for Retrieval of Scientific Publications
At present, academic search engines rely heavily on keyword-based search approaches for the retrieval of relevant scientific literature. This approach has been both efficient and effective in retrieving literature annotated with the same or similar keywords as those contained in a particular search query. In some cases, however, this approach can lead to undesirable search results, as keyword-based search insufficiently account for semantic nuances of individual search terms. This may result in search results containing records that are irrelevant to the search query, or in records annotated with syntactically different but semantically similar keywords not being found. All of this complicates the retrieval of relevant literature, which has an overall negative impact on the productivity and efficiency of the research process as well.
The objective of this thesis is to develop a semantic approach to query expansion for scientific literature retrieval. First, the challenges and limitations of keyword-based approaches are to be outlined using a suitable motivational scenario. The state of the art of query expansion has to be analyzed, with a particular focus on retrieval approaches for scientific literature. Existing solution approaches must be categorized and evaluated against appropriate requirements. A concept for semantic query expansion has to be developed. The feasibility of the concept has to be demonstrated by means of a prototypical implementation and assessed in an adequate evaluation.