Fgram-Tree: An Index Structure Based on Feature Grams for String Approximate Search

摘要

String approximate search is widely used in many areas. Indexing is no doubt a feasible way for efficient approximate string searching. However, the existing index structures have a common weakness that they do not obey the nature of the index which is a function by mapping different data to different index items, similar data to similar index items, in order to query easily. In this paper, we propose a new type of string indexing structure called Fgram-Tree, which is based on feature grams to build itself and filter strings. It obeys the two maps by placing similar strings into the same node, different strings into different nodes that could greatly improve the efficiency of index. Our index is able to support for different types of search. Compared to other index, it provides high scalability and fast response time.