A Study on Top-K algorithm for Large Unordered Tree Databases

2014 Fiscal Year Final Research Report

• Project/Area Number: 24650042
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: Media informatics/Database
• Research Institution: National Institute of Informatics
• Principal Investigator: TAKASU Atsuhiro 国立情報学研究所, コンテンツ科学研究系, 教授 (90216648)
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: 木構造データ検索 / トップＫ検索 / インデキシング

Outline of Final Research Achievements

Trees are used for representing and processing various data such as XML documents and mathematical formulas. We studied efficient tree matching and retrieval algorithms. This study focuses on the algorithms for unordered trees that generally require high computation cost. We first proposed an unordered tree matching algorithm that is especially effective for narrow trees and developed a program that can calculate the similarity of mid sized trees within reasonable processing time. For processing large tree databases, we developed efficient indices that can detect candidate trees from the database. For the case that tree structure is important for retrieval, we made a metric space-based index that converts each tree to a feature vector then makes a metric space for the vectors. Then, we apply a pivot-based indexing technique.

Free Research Field

情報工学