Key Value Storage on Higher Dimentional Data

An Efficient storage system for Higher Dimentional Data .This research work has done as my undergraduate thesis

Overview

The Generalized Two-dimensional Array (G2A) can be used to represent an n-dimensional (nD) array. By mapping the higher-dimensional data into a 2D array, the index (x, y) of each element of that 2D array is used to generate the corresponding key of the record. After storing the keys in a bin-based file system, we design three query paradigms to search for values.

• Exact Matching
• Single key Query
• Range Key Query

Build

To create the keyvalues for the dataset, runcreate.cpp

The number of rows for which keyvalues will be generated can be varied by changing the number in line 59 of create.cpp file

while(fgets(line, sizeof line, fp) != NULL && cnt<10000)

Here, the system will create key values for the first 10,000 rows of the dataset.csv file and store the key values in different bin files.

Exact Matching

Run exact match.cpp

Input the values of a row of the csv file to check whether it exists

For instance, to check for the 7th row, input will be

106	   09:04.1	400	  MAGASIN	   11/3/2014    0:00 	 4:18:44   PM

colunm id is excluded as it is a primary key

Single key Query

For Single Key Query, run singlekeysearch.cpp

To find the relationships of a single value, enter the column number and one value. For example, to determine the relationship of data 106 in column 1, the input will be

1       106	   

Range Key Query

For Range Key Query, run rangevaluesearch.cpp

Only the first column, this feature has been implemented. Input a range of the values in column 1, like 106 and 110, to find all the relations between the range.