Graph Databases for Information Retrieval

Graph models have been deployed in the context of information retrieval for many years. Computations involving the graph structure are often separated from computations related to the base ranking. In recent years, graph data management has been a topic of interest in database research. We propose to deploy graph database management systems to implement existing and novel graph-based models for information retrieval. For this a unifying mapping from a graph query language to graph based retrieval models needs to be developed; extending standard graph database operations with functionality for keyword search. We also investigate how data structures and algorithms for ranking should change in presence of continuous database updates. We want to investigate how temporal decay can affect ranking when data is continuously updated. Finally, can databases be deployed for efficient two-stage retrieval approaches?