Reviewer: Ben Wun
Date: 9-22-2005
How would you rate this paper, relative to others we have read? top 25%, but not top 10%
How would you rate your knowledge of the topic of this paper? novice
What problem or issue does the paper address? Why is it important?
This paper addressed the need for developing a peer to peer searching method that is decentralized, accurate, scalable. This is important because centralized methods have a single point of failure, flooding protocols use too much bandwidth, and most current methods use a fairly simple keyword search that does not effectively rank results.
What are the main contributions of the paper and why are they important?
The authors combined LSI with CAN to produce a scalable, peer to peer IR system called psearch, which uses ranking algorithms to sort results.
How significant are these contributions relative to previous work?
This contribution seems fairly signifigant, since it combines several important techniques for the first time. The authors claim that theirs is the first peer to peer search protocol that is distributed, accurate, and takes into account semantic information in ranking results. Their approach is apparently very scalable, which becomes important as the amount of information that needs to be searched continues to grow.
Give detailed comments justifying your view of the paper.
The authors demonstrate that psearch can be very accurate while visiting relatively few nodes in the network, and with relatively little communication overhead, and demonstrate tradeoffs for performance and storage space. Testing showed psearch should scale well with system size, corpus size, and number of returned documents. While not introducing any particularly new innovation to this area, this paper is important because it shows that combining various approaches leads to a better system.