Improvement in XML Keyword Search and Ranking for Data Analytics

Abstract

The success of web search engine for an ordinary user (Initially, search engine requires very precise query which only expert can write.) motivates the search engine for XML database. XML-based search engine requires DOM parser to parse the XML database. DOM parser produces a tree, which developed only in main memory. But generally XML database is larger than the main memory. Therefore, DOM parser has a disadvantage in case of large database. Instead of using DOM parser, Sax parser is used. SAX parser parses the XML file character by character. Means no requirement of the whole file in main memory, and unlikely DOM parser SAX parser requires no tree. SAX parser consumes less time than DOM Parser also. Searching take a lot of time by hitting the database again and again to fetch the same or recently used data. The solution is a simple cache memory. Cache memory is developed by storing recently used data into hashmap because hash map provides the O(1) search time complexity. Ranking use only use IDF*TF score to calculate the result. But this algorithm does not provide the best ranking. Ranking using cosine similarity algorithm is a better approach. (Basically, Cosine algorithm is used to find similarity between two documents.)