Keyword Search in External Memory Graph

Keyword search over relational and XML data has grown in popularity since the advent of Web search engines. Keyword search over relational data is significantly different from web search as the required information is often split across multiple tables as a result of normalization. The algorithms and techniques that are applied to databases, thus produce answer trees from the data graph as opposed to answer nodes produced by Web search engines. BANKS and some other systems enable keyword-based search over relational databases. Though the algorithms and heuristics used in these systems are highly efficient and have been tuned to give the best results, most of these systems assume the presence of the entire data in memory which may not be the case for standard databases. Also, the search algorithms used by such systems may explore a large portion of the graph before finding an answer. Through this report, we aim to study some of the external memory techniques that can be applied to such systems and thus, improve the time and space complexity of the algorithms, keeping in mind and trying to restrict the number of I/O operations taking place.