Text Mining with Graph Databases: Traversal of Persisted Token-level Representations for Flexible On-demand Processing

摘要

This contribution introduces a new text modelling paradigm that allows for on-demand Text Mining approaches without the need for pre-calculated statistics or data structures other than generic edge-typed adjacency information in a graph database. A simple use cases is presented, for which these new text storing and querying methods can prove beneficial. Furthermore a prototypical implementation is presented and analyzed, showing an adequate runtime behavior for interactive queries.