A Comparative Studies of Automatic Query Formulation in Full-Text Database Search of Chinese Digital Humanities

Chengxi Yan, Tzu-Yi Ho, Jun Wang

Abstract: Query gap is a very serious problem for the full-text database search in the domain of Chinese digital humanities (CDH). General CDH search systems are focused on the improvement of data quality, which ignore the connection between users’ search intents and the system response. We design a two-phase procedure for comparative analysis of the effectiveness of different automatic query formulation in pre-defined tasks, including a prototype system test and a questionnaire-based user study. The experiment shows that compared to query suggestion and query reformulation, query expansion is the most effective automatic query formulation technology with a strong robust performance for user satisfaction, which means it is not sensitive to the types of search intent. The effectiveness of query reformulation and the hybrid methods are limited while query suggestion performs worse in the task for diverse search intent. These findings are believed to be helpful to the reduction of query gap in the full-text database search of Chinese digital humanities, which will foster the development of this field.