@article {12297, title = {Efficient peer-to-peer namespace searches}, journal = {University of Maryland, College Park, MD, Tech. Rep. CSTR-4568}, year = {2004}, month = {2004///}, abstract = {In this paper we describe new methods for efficient and exact search (keyword and full-text) in distributed namespaces. Our methods can be used in conjunction with existing distributed lookup schemes, such as Distributed Hash Tables, and distributed directories. We describe how indexes for implementing distributed searches can be efficiently created, located, and stored. We describe techniques for creating approximate indexes that can be used to bound the space requirement at individual hosts; such techniques are particularly useful for full-text searches that may require a very large number of individual indexes to be created and maintained. Our methods use a new distributed data structure called the view tree. View trees can be used to efficiently cache and locate results from prior queries. We describe how view trees are created, and maintained. We present experimental results, using large namespaces and realistic data, showing that the techniques introduced in this paper can reduce search overheads (both network and processing costs) by more than an order of magnitude.}, author = {Gopalakrishnan,V. and Bhattacharjee, Bobby and Chawathe,S. and Keleher,P.} }