### Citations

4663 | The anatomy of a large-scale hypertextual Web search engine
- BRIN, PAGE
- 1998
(Show Context)
Citation Context ...ank. Hilltop may perform poorly if adequate expert documents are not available for a specic query. (d) SALSA SALSA [30] computes ranks of web pages combining the approaches of HITS [28] and PageRank =-=[15]-=-. For a given query, SALSA computes a set of pages using a search engine (such as Alta-Vista) similar to HITS, which is called base set. From the base set, another set of pages (super set) is identie... |

3624 | Authoritative sources in a hyperlinked environment
- KLEINBERG
- 1999
(Show Context)
Citation Context ...ions such as the inter-connection/linkage between hosts, domains, servers' network addresses or directories to compute approximate PageRank values fast. (b) HITS HITS (Hypertext Induced Topic Search) =-=[28]-=- is another well-known ranking algorithm based on link-structure analysis. HITS employs two scores for each page: hub score and authority score. The computation of these scores is also an iterative pr... |

3264 | The PageRank Citation Ranking: Bringing Order to the Web
- Page, Brin, et al.
- 1998
(Show Context)
Citation Context ...entralized approaches in the literature for ranking Web search results. (a) PageRank and variations (i) Original PageRank The most popular and eective link structure analysis algorithm is PageR8 ank =-=[38]-=-. Google uses PageRank to compute the authority of each crawled webpage. PageRank of a particular webpage A is computed as follows: PR(A) = (1 d) + d i=nX i=1 PR(Bi) Links(Bi) (2.2) Here, d is a damp... |

809 | Chord: A scalable peer-to-peer lookup protocol for internet applications
- Stoica, Morris, et al.
(Show Context)
Citation Context ...e PageRank and keyword relevance. 14 2.5 Web search engines This section presents existing decentralized Web search engines and compares them with DEWS. (a) MINERVA MINERVA [11] is a DHT-based (Chord =-=[45]-=-) decentralized Web search engine. In MINERVA, every peer is autonomous and maintains a local index. Each peer acts like a crawler and posts (using DHT) a small amount of metadata corresponding to the... |

443 | Probabilistic counting algorithms for data base applications
- FLAJOLET, N
- 1985
(Show Context)
Citation Context ... peers from each PeerList using a distributed top-k algorithm [36] so that all the selected peers can be queried in parallel. Global document frequency gdf is computed using the Hash sketch technique =-=[20]-=-. In that approach, every peer includes a hash sketch representing its index list for the respective term when publishing its (term-specic) Post, so that a directory peer can compute an estimate of g... |

409 | Scaling personalized Web search
- Jeh, Widom
- 2003
(Show Context)
Citation Context ...ferences are considered in PageRank by introducing a preference vector (represents the probabilities of the preferred pages) in Equation 2.2. PageRank with user preferences can be computed as follows =-=[26]-=-: PR(A) = (1 d) k=mX k=1 Pk + d i=nX i=1 PR(Bi) Links(Bi) (2.3) Here, m is the number of preferred pages and Pk is the probability of surng the kth preferred page from any page (dened in the prefer... |

317 | The stochastic approach for link-structure analysis
- Lempel, Moran
- 1999
(Show Context)
Citation Context ... may provide better query specic results compared to HITS and topic-sensitive PageRank. Hilltop may perform poorly if adequate expert documents are not available for a specic query. (d) SALSA SALSA =-=[30]-=- computes ranks of web pages combining the approaches of HITS [28] and PageRank [15]. For a given query, SALSA computes a set of pages using a search engine (such as Alta-Vista) similar to HITS, which... |

237 | Topic-Sensitive Pagerank: A Context-Sensitive Ranking Algorithm for Web Search
- Haveliwala
- 2003
(Show Context)
Citation Context ...ageRank [38] does not consider query contexts for ranking the query results. Topics of the queries are classied into several types (e.g., 16 categories from the Open Directory Project (ODP) [41]) in =-=[25]-=- to incorporate the query contexts in PageRank. PageRank vectors for each of the categories are computed o-line. During query processing, user queries are classied into the specic categories and Pa... |

225 | OpenDHT: A public DHT service and its uses.
- Rhea, Dodfrey, et al.
- 2005
(Show Context)
Citation Context ...o the users. During the user registration with CHORA, a summary including URL of the computer, location, bandwidth, and a set of related keywords on that computer is computed and stored using openDHT =-=[42]-=-. CHORA aggregates users browsing history by computing click graphs. Click graph organizes user webpages based on the connectivity implied by their clicks and summary statistics describing their inter... |

158 | Exploiting the block structure of the Web for computing PageRank.
- Kamvar, Haveliwala, et al.
- 2003
(Show Context)
Citation Context ...utton. Parallel PageRank approaches (such as [29], PETSc PageRank [22], and MIKElab PageRank [34]) compute PageRank in parallel to converge quickly. Approximate PageRank algorithms (such as BlockRank =-=[27]-=-, the U-Model [16], and HostRank/DirRank [19]) use higher-level formations such as the inter-connection/linkage between hosts, domains, servers' network addresses or directories to compute approximate... |

156 | Letor: Benchmark dataset for research on learning to rank for information retrieval.
- Liu, Xu, et al.
- 2007
(Show Context)
Citation Context ...n a distributed manner incurring low network and storage overhead. The concepts presented in this work have been validated through extensive simulation results. We used the standard LETOR 3.0 dataset =-=[31]-=- to drive input of our simulation. The organization of the rest of this chapter is as follows. The demand for a decentralized Web search engine and requirements are presented in Section 1.1. 1 Section... |

155 | P-Grid: a selforganizing structured P2P system,”
- Aberer, Cudre-Mauroux, et al.
- 2003
(Show Context)
Citation Context ...s that itsoods the queries in its Gnutella [1] network which requires a large network overhead for query processing. (e) AlvisPeers AlvisPeers [33] is a full-text P2P retrieval engine. It uses P-Grid =-=[8]-=- as the underlying network. The framework of this system has three layers: a) DHT layer which stores global index, b) HDK (Highly Discriminative Keys) layer for building the key vocabulary and corresp... |

114 | Ranking the web frontier,
- Eiron, McCurley, et al.
- 2004
(Show Context)
Citation Context ...[29], PETSc PageRank [22], and MIKElab PageRank [34]) compute PageRank in parallel to converge quickly. Approximate PageRank algorithms (such as BlockRank [27], the U-Model [16], and HostRank/DirRank =-=[19]-=-) use higher-level formations such as the inter-connection/linkage between hosts, domains, servers' network addresses or directories to compute approximate PageRank values fast. (b) HITS HITS (Hyperte... |

100 | ODISSEA: A Peer-to-Peer Architecture for Scalable Web Search and Information Retrieval.
- Suel
- 2003
(Show Context)
Citation Context ...NERVA only employs document frequencies, b) DEWS providessexible searching which may not be possible in the Chord overlay in MINERVA. (b) ODISSEA ODISSEA (Open DIStributed Search Engine Architecture) =-=[46]-=- was a proposal for a P2P search engine for dierent applications including searching P2P networks, large intra-net environment and the Web. This system proposed two tiers: lower layer and upper layer... |

97 | Klee: A framework for distributed top-k query algorithms.
- Michel, Triantafillou, et al.
- 2005
(Show Context)
Citation Context ...m. If a query is initiated in a peer, it retrieves the PeerList for all the query terms by DHT lookups. It selects and contacts with top-k peers from each PeerList using a distributed top-k algorithm =-=[36]-=- so that all the selected peers can be queried in parallel. Global document frequency gdf is computed using the Hash sketch technique [20]. In that approach, every peer includes a hash sketch represen... |

65 |
Space/time tradeos in in hash coding with allowable errors
- Bloom
- 1970
(Show Context)
Citation Context .... Suppose, peer X2 forwards the message to peer X23. Finally, peer X23 can route the message to peer Y since peer X23 will have a routing link for Y = X235 = X23 g5. 2.6.5 Bloom Filter A Bloomslter =-=[13]-=- is a space-ecient probabilistic data structure used to represent a set. Bloomslters support set membership test operations with a small probability of false (erroneous) positives. An empty Bloomslte... |

53 | Minerva: Collaborative p2p search.
- Bender, Michel, et al.
- 2005
(Show Context)
Citation Context ... the websites using both the PageRank and keyword relevance. 14 2.5 Web search engines This section presents existing decentralized Web search engines and compares them with DEWS. (a) MINERVA MINERVA =-=[11]-=- is a DHT-based (Chord [45]) decentralized Web search engine. In MINERVA, every peer is autonomous and maintains a local index. Each peer acts like a crawler and posts (using DHT) a small amount of me... |

53 | Computing pagerank in a distributed internet search system.
- Wang, DeWitt
- 2004
(Show Context)
Citation Context ... PageRank but in a decentralized manner and compute keyword relevance to rank the search results. DEWS providessexible searching and incremental retrieval which are not oered by JXP. (d) Wang et al: =-=[47]-=- proposed a distributed ranking approach where webservers crawl and store only a portion of the Web. Links between webpages stored on dierent servers are discarded, which restricts each server to a p... |

50 | Distributed pagerank for p2p systems.
- Sankaralingam, Sethumadhavan, et al.
- 2003
(Show Context)
Citation Context ...tion presents existing decentralized approaches for ranking Web search results and compares them with our proposed approach DEWS. (a) Sankaralingam et al: proposed a distributed PageRank algorithm in =-=[43]-=- for ranking (HTML) documents available in P2P networks. This approach works with both DHT and non-DHT based P2P networks. It is assumed that documents are pointed by other documents as webpages in th... |

48 |
Efficient PageRank approximation via graph aggregation.
- Broder, Lempel, et al.
- 2006
(Show Context)
Citation Context ...geRank approaches (such as [29], PETSc PageRank [22], and MIKElab PageRank [34]) compute PageRank in parallel to converge quickly. Approximate PageRank algorithms (such as BlockRank [27], the U-Model =-=[16]-=-, and HostRank/DirRank [19]) use higher-level formations such as the inter-connection/linkage between hosts, domains, servers' network addresses or directories to compute approximate PageRank values f... |

34 | Fast Parallel PageRank: A Linear System Approach
- Gleich, Zhukov, et al.
- 2004
(Show Context)
Citation Context ...utton. BackRank ([14], [35]) modies the original PageRank algorithm by adding the possibility of return to the earlier page by back button. Parallel PageRank approaches (such as [29], PETSc PageRank =-=[22]-=-, and MIKElab PageRank [34]) compute PageRank in parallel to converge quickly. Approximate PageRank algorithms (such as BlockRank [27], the U-Model [16], and HostRank/DirRank [19]) use higher-level fo... |

25 | List decoding of q-ary Reed-Muller codes.
- Pellikaan, Wu
- 2004
(Show Context)
Citation Context ...C and Hamming distance from x to each Xi is at most as follows: X(x) = fXijXi 2 C ^ d(Xi; x) g In literature, there are few sophisticated list decoding algorithms including [18], [23], [21], and =-=[40]-=-. Algorithm 1 presents a straight-forward list decoding algorithm, which computes a list of codewords upon receiving a binary pattern where all the codewords are within a pre-specied Hamming distance... |

24 | Efficient and decentralized pagerank approximation in a peer-to-peer web search network. - Parreira, Donato, et al. - 2006 |

22 | Distributed page ranking in structured p2p networks.
- Shi, Yu, et al.
- 2003
(Show Context)
Citation Context ... searching; (d) the concept of incremental retrieval is dierent, and (e) similar to this approach, DEWS caches other peers' addresses as `soft-links'. (b) Shi et al: proposed Open System PageRank in =-=[44]-=- based on structured P2P networks where each peer can communicate and view other peers' webpages. In this approach, webpages are divided into pagegroups using hash code of the websites. If the system ... |

17 | Efficient Parallel Computation of PageRank
- Kohlschütter, Chirita, et al.
- 2006
(Show Context)
Citation Context ...webpages using back button. BackRank ([14], [35]) modies the original PageRank algorithm by adding the possibility of return to the earlier page by back button. Parallel PageRank approaches (such as =-=[29]-=-, PETSc PageRank [22], and MIKElab PageRank [34]) compute PageRank in parallel to converge quickly. Approximate PageRank algorithms (such as BlockRank [27], the U-Model [16], and HostRank/DirRank [19]... |

16 | Framework for decentralized ranking in web information retrieval,”
- Aberer, Wu
- 2003
(Show Context)
Citation Context ... Webpages. This approach computes PageRank in three steps: a) computation of siterank, b) computation of local rankings of webpages, c) combination of the ranking scores using the algebra specied in =-=[9]-=-. In this approach, a sitegraph is dened using the collection of websites and their internal links. The assumptions regarding the computation of sitegraph are as follows: a) the size of the whole Web... |

15 |
Hilltop: A Search Engine: Based on Expert Documents,
- Bharat, Mihaila
- 1999
(Show Context)
Citation Context ...HITS performs ranking considering query contexts. HITS requires more query resolution time than PageRank as HITS computes the root and base sets during the query evaluation phase. (c) Hilltop Hilltop =-=[12]-=- maintains a set of expert documents, which allow to provide query specic pages in search results. Expert documents are the subset of the crawled pages, which are topic specic and have links to many... |

11 |
The eect of the back button in a random walk: application for pagerank
- Mathieu, Bouklit
- 2004
(Show Context)
Citation Context ... to the clusters, which is very similar to the topic sensistive PageRank. Original PageRank algorithm does not consider the possibility of browsing visited webpages using back button. BackRank ([14], =-=[35]-=-) modies the original PageRank algorithm by adding the possibility of return to the earlier page by back button. Parallel PageRank approaches (such as [29], PETSc PageRank [22], and MIKElab PageRank ... |

10 |
An improved list decoding algorithm for the second order reed{muller codes and its applications
- Fourquet, Tavernier
(Show Context)
Citation Context ...here Xi 2 C and Hamming distance from x to each Xi is at most as follows: X(x) = fXijXi 2 C ^ d(Xi; x) g In literature, there are few sophisticated list decoding algorithms including [18], [23], =-=[21]-=-, and [40]. Algorithm 1 presents a straight-forward list decoding algorithm, which computes a list of codewords upon receiving a binary pattern where all the codewords are within a pre-specied Hammin... |

10 | Alvis peers: a scalable full-text peer-to-peer retrieval engine
- Luu, Klemm, et al.
- 2006
(Show Context)
Citation Context ...e of this approach towards a scalable Web search engine is that itsoods the queries in its Gnutella [1] network which requires a large network overhead for query processing. (e) AlvisPeers AlvisPeers =-=[33]-=- is a full-text P2P retrieval engine. It uses P-Grid [8] as the underlying network. The framework of this system has three layers: a) DHT layer which stores global index, b) HDK (Highly Discriminative... |

9 | Using siterank for p2p web retrieval
- Wu, Aberer
(Show Context)
Citation Context ...d servers. In contrast to this algorithm, we compute PageRank on the whole Web and keyword relevance in decentralized manners. We also providesexible searching and incremental retrieval. (e) SiteRank =-=[49]-=- proposed a decentralized system architecture [48] to compute ranks of Webpages. This approach computes PageRank in three steps: a) computation of siterank, b) computation of local rankings of webpage... |

7 | Backrank: an alternative for pagerank
- Bouklit, Mathieu
- 2005
(Show Context)
Citation Context ...eights to the clusters, which is very similar to the topic sensistive PageRank. Original PageRank algorithm does not consider the possibility of browsing visited webpages using back button. BackRank (=-=[14]-=-, [35]) modies the original PageRank algorithm by adding the possibility of return to the earlier page by back button. Parallel PageRank approaches (such as [29], PETSc PageRank [22], and MIKElab Pag... |

7 |
Parallel PageRank Computation on a Gigabit PC Cluster.
- Manaskasemsak, Rungsawang
- 2004
(Show Context)
Citation Context ...) modies the original PageRank algorithm by adding the possibility of return to the earlier page by back button. Parallel PageRank approaches (such as [29], PETSc PageRank [22], and MIKElab PageRank =-=[34]-=-) compute PageRank in parallel to converge quickly. Approximate PageRank algorithms (such as BlockRank [27], the U-Model [16], and HostRank/DirRank [19]) use higher-level formations such as the inter-... |

6 |
List decoding of reed-muller codes up to the johnson bound with almost linear complexity
- Dumer, Kabatiansky, et al.
- 2006
(Show Context)
Citation Context ... : : ; Xmg where Xi 2 C and Hamming distance from x to each Xi is at most as follows: X(x) = fXijXi 2 C ^ d(Xi; x) g In literature, there are few sophisticated list decoding algorithms including =-=[18]-=-, [23], [21], and [40]. Algorithm 1 presents a straight-forward list decoding algorithm, which computes a list of codewords upon receiving a binary pattern where all the codewords are within a pre-spe... |

6 |
Weighted pagerank: cluster-related weights
- Nemirovsky
- 2008
(Show Context)
Citation Context ...e PageRank of page A as a category c, mc is the number of predened pages for category c, and P ck is the probability of k th page in category c. 9 (iv) Other variations of PageRank Weighted PageRank =-=[37]-=- groups the URLs into clusters and assign weights to the clusters, which is very similar to the topic sensistive PageRank. Original PageRank algorithm does not consider the possibility of browsing vis... |

5 | Chora: Expert-based p2p web search
- Gylfason, Khan, et al.
- 2008
(Show Context)
Citation Context ...them. Authors proposed a Bloomslter based protocol for optimizing query execution plans. In contrast with ODISSEA, we present a complete Web search engine with specic implementation. (c) CHORA CHORA =-=[24]-=- is not a standalone search engine rather it enhances the current centralized search engines to incorporate the users' browsing history into search results. The framework of CHORA consists of two sear... |

5 | Towards a decentralized search architecture for the web and p2p systems
- Wu
- 2003
(Show Context)
Citation Context ...te PageRank on the whole Web and keyword relevance in decentralized manners. We also providesexible searching and incremental retrieval. (e) SiteRank [49] proposed a decentralized system architecture =-=[48]-=- to compute ranks of Webpages. This approach computes PageRank in three steps: a) computation of siterank, b) computation of local rankings of webpages, c) combination of the ranking scores using the ... |

4 |
Plexus: a scalable peer-to-peer protocol enabling efficient subset search,”
- Ahmed, Boutaba
- 2009
(Show Context)
Citation Context ...o the advertisement keyword. The concept of keyword matching using list decoding is discussed in Section 2.6.4. Figure 4.2(b) shows the eectiveness of list decoding technique. 2.6.4 Plexus In Plexus =-=[10]-=-, keywords are mapped to patterns (or bit-vectors) and a Hamming distance based routing technique derived from the theory of Linear Binary Codes is used. The keyword to pattern mapping process retains... |

3 | Towards a fully distributed p2p web search engine
- Zhou, Li, et al.
- 2004
(Show Context)
Citation Context ... search results returned by the traditional search engines. In contrast, our motivation behind DEWS is to develop a standalone decentralized Web search engine. (d) COOPER The motivation behind COOPER =-=[50]-=- is similar to CHORA. It works with centralized search engines and incorporates users searching experiences on the search results in a Peer-to-Peer fashion . Instead of using PageRank, COOPER proposes... |

2 |
A distributed ranking strategy in peer-to-peer based information retrieval systems
- Lu, Ling, et al.
- 2004
(Show Context)
Citation Context ...y searching the local repository of that node. However, it is not trivial to compute U in a purely decentralized setup. We use the total number of indexed URLs in a node in place of U as advocated in =-=[32]-=-. 3.6 Web indexing and searching We assume that machines running one or more webservers provide search functionalities and index their hosted websites. 3.6.1 Website indexing We assume that an AI (Art... |

1 |
Coding Theory;The Essentials
- Homan, Leonard, et al.
(Show Context)
Citation Context ... C can be represented as < n; k; d > where n is the number of bits (0 or 1) in C, k is the dimension of the generator matrix of C, and d is the minimum Hamming distance between any two codewords in C =-=[17]-=-. All the codewords of a particular linear binary code can be represented by a minimal set of codewords, which is known as generator matrix. A generator matrix GC of a linear binary code C has k rows.... |

1 |
List-decoding reed-muller codes over small
- Gopalan, Klivans, et al.
- 2008
(Show Context)
Citation Context ... Xmg where Xi 2 C and Hamming distance from x to each Xi is at most as follows: X(x) = fXijXi 2 C ^ d(Xi; x) g In literature, there are few sophisticated list decoding algorithms including [18], =-=[23]-=-, [21], and [40]. Algorithm 1 presents a straight-forward list decoding algorithm, which computes a list of codewords upon receiving a binary pattern where all the codewords are within a pre-specied ... |