A. Kronfol. FASD: a fault-tolerant, adaptive, scalable, distributed search engine. June 2002. Unpublished manuscript.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....is that all or most nodes need to be contacted for most queries, and thus such schemes are unlikely to scale beyond a few hundred nodes. There have been attempts to overcome this issue by routing queries only to those nodes that are likely to have good results or are in the vicinity [27, 43, 16]. However, we do not believe that this approach will scale at all if result quality is a major concern, since document collections are simply not naturally clustered in a way that allows queries to be routed to only a small fraction of the nodes. This is certainly the case for the current web, ....

.... that the problem of full text search on terabyte size collections is different from that on smaller collections or on systems that only index titles and keywords for multimedia objects (e.g. mp3 files) Some recent work on text search in P2P systems with local index organization appears in [16, 27, 43, 47]. As explained, the global index organization is one of the aspects that distinguish our system from others. Another very different approach to distributed search is taken by systems such as JXTA [34] STARTS [21] and the Z39.50 standard [36] which are mainly concerned with issues of combining ....

A. Kronfol. FASD: a fault-tolerant, adaptive, scalable, distributed search engine. June 2002. Unpublished manuscript.

....[3] system merges the WHERE and HOW stages and relies upon a separate search layer to implement the WHAT stage. Merging stages may be necessary in some cases, however to the extent that they are separable they can be implemented by entirely different systems. For example, one might use FASD [12] to identify a file from keyword meta data, use Chord [20] to work out the location of the file itself, and then BitTorrent [5] to actually download it. Distributed Hashtables such as Chord and CAN [14] with their bounds on path length and or connectivity as network size increases, seem well ....

....such as MojoNation [23] now Mnet) which provides all three stages in one package. Recently a number of systems have been developed that try to provide sophisticated meta data search. One possible P2P meta data approach is to try and use Chord to store keyword document relations. Kronfol [12] suggests that under this scheme popular query terms would drive excessive traffic to certain nodes. As an alternative Kronfol describes and simulates FASD, which adds keyword searching to the Freenet system by inserting meta data keys that include the TFIDF rankings of keywords in Freenet ....

Kronfol, A.Z.: FASD: A fault-tolerant, Adaptive Scalable, Distributed Search Engine. Princeton University Technical Report. (2002). http://www.cs.princeton.edu/~akronfol/fasd/

....modifications to existing systems , have come to light in the last year or so, perhaps indicating a desire to move beyond the search capabilities of the first generation of P2P software . One possible P2P meta data approach is to try and use Chord to store keyworddocument relations [19] Kronfol [10] suggests that under this scheme popular query terms would drive excessive traffic to certain nodes. As an alternative Kronfol describes and simulates FASD, which adds keyword searching to the Freenet system by inserting meta data keys that include the TFIDF rankings of keywords in Freenet ....

Kronfol, A.Z.: FASD: A fault-tolerant, Adaptive Scalable, Distributed Search Engine. Princeton University Technical Report. (2002). http://www.cs.princeton.edu/~akronfol/fasd/

....would map individual words instead of object identi ers as keys. In fact, except for very speci c uses, such as in Freenet [4] where the object identi er must be publicized outside the system for con dentiality purposes, word search is a basic requirement for a usable document indexing system [8]. Since words typically follow a Zipf distribution [5, 2] depicted in the line on Figures 2 and 3) with a small set of words present in many description phrases (e.g. two, the, mp3) it is easy to conclude how the key image size would be a ected for those words. In the following sections we ....

Amr Z. Kronfol. FASD: A Fault-tolerant, Adaptive, Scalable, Distributed Search Engine. PhD thesis, May 2002.

No context found.

A. Kronfol. FASD: a fault-tolerant, adaptive, scalable, distributed search engine. June 2002. Unpublished manuscript.

....is that all or most nodes need to be contacted for most queries, and thus such schemes are unlikely to scale beyond a few hundred nodes. There have been attempts to overcome this issue by routing queries only to those nodes that are likely to have good results or are in the vicinity [11, 18, 29]. However, we do not believe that this approach will scale if result quality is a major concern, since document collections are simply not naturally clustered in a way that allows queries to be routed to only a small fraction of the nodes. This is certainly the case for the current web, where a ....

.... that the problem of full text search on terabyte size collections is different from that on smaller collections or on systems that only index titles and keywords for multimedia objects (e.g. mp3 files) Some recent work on text search in P2P systems with local index organization appears in [11, 18, 29, 33]. As explained, the global index organization is one of the aspects that distinguish our system from others. Another very different approach to distributed search is taken by systems such as JXTA [22] STARTS [16] and the Z39.50 standard [23] which are mainly concerned with issues of combining ....

A. Kronfol. FASD: a fault-tolerant, adaptive, scalable, distributed search engine. June 2002. Unpublished manuscript.

....The main problem with local indexes is that all or most nodes are contacted for most queries, and thus they are unlikely to scale beyond a few hundred nodes. There have been attempts to overcome this issue by routing queries only to nodes likely to have good results or those in the vicinity [4, 6, 13]. However, we do not believe that this approach will work well if result quality is a major concern. To see this, consider the current web, where an approach based on local indexes at each site would be either extremely inefficient or give very poor answers. With a global index, on the other ....

A. Kronfol. FASD: a fault-tolerant, adaptive, scalable, distributed search engine. June 2002. Unpublished manuscript.

No context found.

A. Kronfol. FASD: a fault-tolerant, adaptive, scalable, distributed search engine. June 2002. Unpublished manuscript.

No context found.

A. Z. Kronfol. FASD: A Fault-tolerant, Adaptive, Scalable, Distributed Search Engine. Technical report, Princeton, 2002.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC