| Presentation | 2014/6/18 Performance Evaluation of Parallel Processing for Geometric Suffix Tree Yoshifumi Takahashi, Keiichi Tamura, Susumu Kuroki, Hajime Kitakami, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Geometric suffix trees stored on a disk can be used as indices to achieve a high-speed similarity search on large-scale, 3-D protein structure databases. However, due to an increasing number of data records in the protein structure database, a large amount of time must be taken to construct the index, greatly affecting the search speed. This paper proposes a parallel processing method using a master worker and distributed worker models in order to parallelize both the construction algorithm of the index and the similarity search algorithm using the index constructed in the protein structure database. The authors focus on the viewpoint that the existing construction method of the geometric suffix tree is not suitable for direct parallelization. Therefore, beforehand, the authors changed the existing incremental construction algorithm of the geometric suffix tree into a top-down algorithm to process all of the data together in the database. Moreover, the authors changed the existing buffer control method into an adaptive method for parallel processing. In order to parallelize both the construction and the similarity search, both data partition and task partition methods are applied to a given problem, such as the construction and similarity search, and sub-problems generated from the given problem are executed in parallel using each worker model. The performance evaluation experiments resulted as follows: in the parallel construction of the index, the data partition method was better than the task partition method; in the parallel similarity search using the index, the task partition method was better than the data partition method. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | |
| Paper # | Vol.2014-MPS-98 No.21,Vol.2014-BIO-38 No.21 |
| Date of Issue |
| Conference Information | |
| Committee | IBISML |
|---|---|
| Conference Date | 2014/6/18(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Information-Based Induction Sciences and Machine Learning (IBISML) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Performance Evaluation of Parallel Processing for Geometric Suffix Tree |
| Sub Title (in English) | |
| Keyword(1) | |
| 1st Author's Name | Yoshifumi Takahashi |
| 1st Author's Affiliation | Graduate School of Information Science, Hiroshima City University() |
| 2nd Author's Name | Keiichi Tamura |
| 2nd Author's Affiliation | Graduate School of Information Science, Hiroshima City University |
| 3rd Author's Name | Susumu Kuroki |
| 3rd Author's Affiliation | Graduate School of Information Science, Hiroshima City University |
| 4th Author's Name | Hajime Kitakami |
| 4th Author's Affiliation | Graduate School of Information Science, Hiroshima City University |
| Date | 2014/6/18 |
| Paper # | Vol.2014-MPS-98 No.21,Vol.2014-BIO-38 No.21 |
| Volume (vol) | vol.114 |
| Number (no) | 105 |
| Page | pp.pp.- |
| #Pages | 8 |
| Date of Issue | |