A Value-Attribute Mapping Method Based on Inverted Index

The emergence of database natural language interface has facilitated database access and search by more and more non-professional users. In the process of natural language transformation, node mapping is the most basic and critical step, especially the mapping of value nodes. However, the method of traversing the database to search value nodes is inefficient. In this paper, we propose a novel "Value-Attribute" mapping method based on hierarchical structure inverted index to quickly search the fields which the value belongs to. First of all, we design a first-level inverted index, creating the mapping between values and attributes. Then in this period, the criteria are proposed to determine whether or not to create inverted index for a field. To optimize the first-level inverted index, we classify the inverted index according to the data type and add the B+ tree data structure. Finally, the experimental results on real medical dataset verify the effectiveness and efficiency of the proposed hierarchical structure inverted index.