Masterarbeit
Leveraging Knowledge Graphs for Long-Tail Query Auto-Completion in Research Data Repositories
Completion
2025/01
Research Area
Intelligent Information Management
Students
Shwetaben Ramanikbhai Khatra
Advisers
Description
Query Auto-Completion (QAC) is a fundamental component of information retrieval systems, assisting users in formulating their queries by providing suggestions for query completion. Traditional query auto-completion approaches rely on rule-based, heuristic or learning-based approaches. While some methods evaluate semantic relationships among search terms using graph-based techniques, the potential of Semantic Web technologies remains underexplored.
In research data repositories, detailed metadata descriptions are typically available, encompassing administrative metadata about authors or creators, associated research areas, keywords, and content-related descriptive metadata. Furthermore, there are relationships between data and their associated authors or creators, as well as among individual datasets. This comprehensive information forms a semantic network that can be represented in a knowledge base. Leveraging such a knowledge base, various metrics can be incorporated into the process of ranking candidates for query completion.
The goal of this work is to utilize the benefits associated with knowledge graphs to enhance query auto-completion for long-tail queries, which are typically more complex and contain more keywords compared to short-tail queries. The unique characteristic of long-tail queries is that they provide contextual information that can be leveraged by semantic query auto-completion approaches. To this end, a requirement analysis must first be carried out, and the state of the art must be investigated. Existing solutions are to be classified and evaluated according to the requirements. Subsequently, a query auto-completion approach for long-tail queries using semantic methods has to be designed and described. The designed approach has to be demonstrated in a prototypical implementation. In addition, the practicability and acceptance of the approach must be assessed in a suitable evaluation.
