Efficient Multidimensional Database Indexing Techniques

Abstract

Conventional multidimensional database indexing systems are efficient in storing the newlinedata at physical level and provide fast accessibility to the stored records as and when newlineuser raises a query seeking information to given keywords as input to the information newlinesystem. A novice user has a tendency to do typographical errors as a result of newlinemisstyped keywords and cognitive errors due to misunderstanding of correct spelling newlineleading to failure in accessing information. This problem arises because traditional newlineindexing systems lack fault-tolerance ability in mapping misspelled keywords to the newlinecorrect records. newline newlineAnother entropy based multidimensional clustering is proposed for generating the newlinemultidimensional cluster tables. This method is based on the statistical structure of newlinelanguage, as successive symbols are not chosen independently but their probabilities newlinedepend on the preceding letters. Dimensional keywords are stored as cluster keys newlinealong with the tagged value generated using the arithmetic coding technique. At the newlinetime of accessing the data value, the new tagged value is computed for the input keyword newline.The tagged interval of input keyword is computed, while this new tagged interval newlinewill either contain required interval of stored cluster key or lie within the tagged newlineinterval of stored cluster key. This process provides fault tolerance ability to multidimensional newlineclustering of table in accessing data when misspelled keys are directly newlinegiven as input to the query process. newline newlineIt has been observed that the results of all the proposed methods were found to newlinemeet the proposed objectives of the thesis. All proposed methods have shown excellent newlinefault-tolerance ability in mapping the misspelled keywords to the records at newlinephysical level. It has been found that qualitatively all the proposed methods are at newlinethe winning position against the conventional indexing method because conventional newlineindexing method does not have fault tolerance ability. Exciting results were found during newlineexperiments performed on multidimensional databases to v