An approximate search engine for structural databases

ABSTRACT

When a person interested in a topic enters a keyword into a Web search engine, the response is nearly instantaneous (and sometimes overwhelming). The impressive speed is due to clever inverted index structures, caching, and a domain-independent knowledge of strings. Our project seeks to construct algorithms, data structures, and software that approach the speed of keyword-based search engines for queries on structural databases.A structural database is one whose data objects include trees, graphs, or a set of interrelated labeled points in two, three, or higher dimensional space. Examples include databases holding (i) protein secondary and tertiary structure, (ii) phylogenetic trees, (iii) neuroanatomical networks, (iv) parse trees, (v) molecular diagrams, and (vi) XML documents. Comparison queries on such databases require solving variants of the graph isomorphism or subisomorphism problems (for which all known algorithms are exponential), so we have explored a large heuristic space.

INDEX TERMS

Primary Classification:
  H. Information Systems
  H.3 INFORMATION STORAGE AND RETRIEVAL

Additional Classification:
  H. Information Systems
  H.3 INFORMATION STORAGE AND RETRIEVAL

General Terms:
Algorithms, Design, Management