Collaborative Research: ASES: An Approximate Search Engine for Structure

本研究项目的目标是使对组合结构（如树和图）的近似查询以快速（即，相对于被搜索对象的数量是次线性的）和有效的方式处理成为可能。最终目标是在文本文档中实现近似关键字搜索的速度。该方法包括：(1)发现在多项式时间内以高质量回答这些np完全问题的最佳启发式方法，以及(2)发现数据结构以快速查询数千或数百万个此类对象。该项目的结果将提供在图和树结构数据中搜索模式的技术和软件。这种搜索引擎的可能应用包括对蛋白质、化合物、神经解剖结构、Web/文本过滤器和XML文档的搜索。本研究开发的技术特别适用于生物信息学和生物计算应用。Shasha教授和Wang教授计划在这些领域与另外五位研究人员合作：Jack Collins，国家癌症研究所从事药物设计的小分子-蛋白质对接；生物学教授兼哈佛大学植物标本室主任Michael Donohue计划将这项研究应用于系统发育树；Bruce Shapiro，国家癌症研究所开发算法和计算系统，用于确定核酸的结构/功能；Cathy Wu，国家生物医学研究基金会从事蛋白质序列分析和分类研究；以及国家癌症研究所的Daniel Zaharevitz，他的目标是通过更好的搜索能力使研究团体更容易获得生物和结构数据。这些合作者有助于激励和验证生物应用的树和图匹配工具和算法开发。http://www.cis.njit.edu/~jason/sigmod.html

The goal of this research project is to make it possible to process approximate queries on combinatorial structures such as trees and graphs in a fast (i.e., sublinear with respect to the number of objects being searched) and efficient manner. The ultimate goal is to be as fast as approximate keyword search in text documents. The approach consists of: (1) discovering the best heuristics for answering these NP-complete problems in polynomial time with high quality, and (2) discovering data structures to make queries on thousands or millions of such objects fast. The results of this project will provide techniques and software to search for patterns among graph and tree structured data. Possible applications of such a search engine include searches among proteins, chemical compounds, neuroanatomical structures, Web/text filters and XML documents. The techniques developed in this research are in particular suitable for bioinformatics and biocomputing applications. Professors Shasha and Wang plan to collaborate with five additional researchers in these areas: Jack Collins, National Cancer Institute working in small molecule-protein docking for drug design; Michael Donohue, Professor of Biology and Director of the Harvard University Herbaria planning to apply this research to phylogenic trees; Bruce Shapiro, National Cancer Institute developing algorithms and computational systems for determining structure/function of nucleic acids; Cathy Wu, National Biomedical Research Foundation doing research in analysis and classification of protein sequences; and Daniel Zaharevitz, National Cancer Institute whose goal is to make biological and structural data more available to the research community via better search capabilities. These collaborators help to motivate and validate the tree and graph matching tools and algorithms development for biological applications.http://www.cis.njit.edu/~jason/sigmod.html