Parallel Data Mining Using Multi-Core Computing
School: Los Alamos High
Area of Science: Computing
Proposal: Proposal: Team 1029
For this year's supercomputing challenge, our team will create a search engine for data storage. We will then try and optimize this engine and compare it with conventional engines. We will run the program for a specific file on a large data storage and time how long the search runs. We will then run the conventional engine for the same file, timing that and seeing which one is faster. We also hope to modify our program for use on the internet. We will also optimize that for search internet use, and possibly leave it there for public use. We will create this program with java, in two separate parts. The first part will be a data indexer, which will speed up the process of the search and provide criteria and/or key words for the actual engine to use. The second part is the search engine itself. In this we will apply the type of search method that we find most appropriate for the situation (breadth-first, depth-first, etc). The data set we will use at first will be small, and used as a control set for the current time. For successive tests, we will add more files into the data set while recording the time taken for each test. After this, we will look at the time change between data sets and attempt to reduce that change by as much as we deem possible. We might be to obtain more data sets from other sources as well because manually creating sufficient numbers of data sets is next to impossible.
Sponsoring Teacher: Wyatt Dumas
Mail the entire Team