Improving Query Performance on Database Free-Text and Semi-Structured Text Columns
The project deals with the better indexing of text fields in the database. It explores an idea of on-line maintenance of an inverted index, which is built for the tokens of the text content in the database. The on-line update of such an index is currently considered impractical. We try to overcome this inverted index update bottleneck with the use of B-tree data structure and the specially designed in-memory buffer. In case of success, this on-line indexing will highly improve the performance of user queries on database text fields. In addition, it may create a viable solution for other fields, such as WEB documents indexing, which require the on-line incorporation of up-to-date changes in the index.