DMCA
Optimizing web search using social annotations (2007)
Venue: | IN: WWW ’07 |
Citations: | 171 - 2 self |
Citations
4669 | The anatomy of a large-scale hypertextual Web search engine
- Brin, Page
- 1998
(Show Context)
Citation Context ...ngsthe quality of web search. Most of them contribute from twosaspects: 1) ordering the web pages according to the querydocument similarity. State-of-the-art techniques include anchorstext generation =-=[21, 28, 34]-=-, metadata extraction [37], linksanalysis [34], and search log mining [10]; 2) ordering the webspages according to their qualities. It is also known as queryindependent ranking, or static ranking. For... |
4017 |
Introduction to modern information retrieval
- Salton, McGill
- 1983
(Show Context)
Citation Context ...g and staticsranking as follows.sSimilarity ranking measures the relevance between a query andsa document. Many models have been proposed to estimate thessimilarity between the query and the document =-=[11]-=-. In modernssearch engines, several methods have been proposed to find newsinformation as additional metadata to enhance the performance ofssimilarity ranking, e.g., document title [37], anchor text [... |
3631 | Authoritative sources in a hyperlinked environment
- Kleinberg
- 1999
(Show Context)
Citation Context ...bspages according to their qualities. It is also known as queryindependent ranking, or static ranking. For a long time, the staticsranking is derived based on link analysis, e.g., PageRank [17],sHITS =-=[15]-=-. Recently, the features of content layout, user clickthroughs etc. are also explored, e.g., fRank[19]. Given a query,sthe retrieved results are ranked based on both page quality andsquery-page simila... |
3268 | The pagerank citation ranking: Bringing order to the web
- Page, Brin, et al.
- 1999
(Show Context)
Citation Context ...ring the webspages according to their qualities. It is also known as queryindependent ranking, or static ranking. For a long time, the staticsranking is derived based on link analysis, e.g., PageRank =-=[17]-=-,sHITS [15]. Recently, the features of content layout, user clickthroughs etc. are also explored, e.g., fRank[19]. Given a query,sthe retrieved results are ranked based on both page quality andsquery-... |
1311 | Optimizing search engines using clickthrough data
- Joachims
- 2002
(Show Context)
Citation Context ...ntegrating the social features into the rankingsalgorithm. In our work, we incorporate both similarity and staticsfeatures exploited from social annotations into the rankingsfunction by using RankSVM =-=[32]-=-.s3.4.2 FeaturessWe divided our feature set into two mutually exclusive categories:squery-page similarity features and page’s static features. Table 1sdescribes each of these feature categories in det... |
532 | Learning to rank using gradient descent
- Burges, Shaked, et al.
- 2005
(Show Context)
Citation Context ...matic) tuning ofsspecific ranking functions. Previous work estimates the weightssthrough regression [26]. Recent work on this ranking problemsattempts to directly optimize the ordering of the objects =-=[3, 22,s32]-=-.sAs discussed in [5], there are generally two ways to utilize thesexplored social features for dynamic ranking of web pages: (a)streating the social actions as independent evidence for rankingsresult... |
512 |
Usage patterns of collaborative tagging systems.
- Golder, Huberman
- 2006
(Show Context)
Citation Context ...lz argued that a folksonomy could be quitesuseful in that it revealed the digital equivalent of “desire lines”.sDesire lines were the foot-worn paths that sometimes appeared insa landscape over time. =-=[27]-=- analyzed the structure of collaborativestagging systems as well as their dynamical aspects. Hotho et al.sproposed Adapted PageRank and FolkRank to find communitiesswithin the folksonomy but have not ... |
466 | Ontologies Are Us: A Unified Model of Social Networks and Semantics”,
- Mika
- 2005
(Show Context)
Citation Context ...otations and capture the preference of web annotators.s2.2 Research on Social AnnotationssExisting research on social annotations includes “folksonomy” [2,s24], visualization [18], emergent semantics =-=[25]-=-, semantic webs[36], enterprise search [23] etc.s“Folksonomy”, a combination of “folk” and “taxonomy”, wassfirst proposed by T. V. Wal in a mailing list [12]. Folksonomyswas further divided into the n... |
413 | Ir evaluation methods for retrieving highly relevant documents
- Järvelin, Kekäläinen
- 2000
(Show Context)
Citation Context ..., (9) where p(j) denotes the precision over the top j results, andsΔr(j) is the change in recall from j-1 to j.sz NDCG at K: NDCG is a retrieval measure devisedsspecifically for web search evaluation =-=[16]-=-. It is well suitedsto web search evaluation as it rewards relevant documentssthat are top-ranked more heavily than those ranked lower.sFor a given query q, the ranked results are examined in astop-do... |
386 | SimRank: A measure of structural-context similarity
- Jeh, Widom
- 2002
(Show Context)
Citation Context ...s equal to MAP (aj, pn). Figures2(b) shows the first iteration’s SSR result of the sample dataswhere CA and CP are set to 1.sThe convergence of the algorithm can be proved in a similarsway as SimRank =-=[9]-=-. For each iteration, the time complexity ofsthe SSR algorithm is O(NA2NP2). Within the data set of oursexperiment, both the annotation and web page similarity matricessare quite sparse and the algori... |
378 |
Improving web search ranking by incorporating user behavior information.
- Agichtein, Brill, et al.
- 2006
(Show Context)
Citation Context ...king functions. Previous work estimates the weightssthrough regression [26]. Recent work on this ranking problemsattempts to directly optimize the ordering of the objects [3, 22,s32].sAs discussed in =-=[5]-=-, there are generally two ways to utilize thesexplored social features for dynamic ranking of web pages: (a)streating the social actions as independent evidence for rankingsresults, and (b) integratin... |
327 | Folksonomies – cooperative classification and communication through shared metadata (2004). Computer Mediated Communication. Available at: www.adammathes.com/academic/computermediated-communication/folksonomies.html (accessed 20
- Mathes
- 2008
(Show Context)
Citation Context ...than 200% in the past nine months [13]. Social annotationssare emergent useful information that can be used in various ways.sSome work has been done on exploring the social annotations forsfolksonomy =-=[2]-=-, visualization [18], semantic web [36], enterprisessearch [23] etc. However, to the best of our knowledge, littleswork has been done on integrating this valuable information intosweb search. How to u... |
164 | Effective site finding using link anchor information.
- Craswell, Hawking, et al.
- 2001
(Show Context)
Citation Context ...ngsthe quality of web search. Most of them contribute from twosaspects: 1) ordering the web pages according to the querydocument similarity. State-of-the-art techniques include anchorstext generation =-=[21, 28, 34]-=-, metadata extraction [37], linksanalysis [34], and search log mining [10]; 2) ordering the webspages according to their qualities. It is also known as queryindependent ranking, or static ranking. For... |
160 | Visualizing tags over time.
- Dubinko
- 2007
(Show Context)
Citation Context ...st nine months [13]. Social annotationssare emergent useful information that can be used in various ways.sSome work has been done on exploring the social annotations forsfolksonomy [2], visualization =-=[18]-=-, semantic web [36], enterprisessearch [23] etc. However, to the best of our knowledge, littleswork has been done on integrating this valuable information intosweb search. How to utilize the annotatio... |
138 |
Exploring social annotations for the semantic web. In:
- Wu, Zhang, et al.
- 2006
(Show Context)
Citation Context .... Social annotationssare emergent useful information that can be used in various ways.sSome work has been done on exploring the social annotations forsfolksonomy [2], visualization [18], semantic web =-=[36]-=-, enterprisessearch [23] etc. However, to the best of our knowledge, littleswork has been done on integrating this valuable information intosweb search. How to utilize the annotations effectively to i... |
108 | Log-linear models for label ranking
- Dekel, Manning, et al.
- 2003
(Show Context)
Citation Context ...matic) tuning ofsspecific ranking functions. Previous work estimates the weightssthrough regression [26]. Recent work on this ranking problemsattempts to directly optimize the ordering of the objects =-=[3, 22,s32]-=-.sAs discussed in [5], there are generally two ways to utilize thesexplored social features for dynamic ranking of web pages: (a)streating the social actions as independent evidence for rankingsresult... |
103 |
Social bookmarking tools (i): A general review. D-Lib Magazine
- Hammond, Hannay, et al.
- 2005
(Show Context)
Citation Context ...hesGoogle’s toolbar API.sSocialPageRanks(SPR)sThe popularity score calculated based onsSocialPageRank algorithm.s4. EXPERIMENTAL RESULTSs4.1 Delicious DatasThere are many social bookmark tools on Web =-=[30]-=-. For thesexperiment, we use the data crawled from Delicious during May,s2006, which consists of 1,736,268 web pages and 269,566sdifferent annotations.sAlthough the annotations from Delicious are easy... |
101 | Support vector learning for ordinal regression
- Herbrich, Graepel, et al.
- 1999
(Show Context)
Citation Context ...nk results by learning a rank function. Many methodsshave been developed for automatic (or semi-automatic) tuning ofsspecific ranking functions. Previous work estimates the weightssthrough regression =-=[26]-=-. Recent work on this ranking problemsattempts to directly optimize the ordering of the objects [3, 22,s32].sAs discussed in [5], there are generally two ways to utilize thesexplored social features f... |
97 | Statistical models for co-occurrence data.
- Hofmann, Puzicha
- 1998
(Show Context)
Citation Context ...efer toseither airplane ticket or concert ticket, and terms with these twosdifferent meanings will be mixed up. In [36], Wu et al. studied thesproblem of annotation ambiguity by using a mixture model =-=[31]-=-;showever, it is not suitable for the web search due to its highscomputational complexity. Some efficient disambiguationsmethods may be required for further improving the performancesof SSR. However, ... |
88 |
Optimizing web search using web click-through data
- Xue, Zeng, et al.
- 2004
(Show Context)
Citation Context ...g the web pages according to the querydocument similarity. State-of-the-art techniques include anchorstext generation [21, 28, 34], metadata extraction [37], linksanalysis [34], and search log mining =-=[10]-=-; 2) ordering the webspages according to their qualities. It is also known as queryindependent ranking, or static ranking. For a long time, the staticsranking is derived based on link analysis, e.g., ... |
64 | Beyond pagerank: machine learning for static ranking.
- Richardson, Prakash, et al.
- 2006
(Show Context)
Citation Context ... For a long time, the staticsranking is derived based on link analysis, e.g., PageRank [17],sHITS [15]. Recently, the features of content layout, user clickthroughs etc. are also explored, e.g., fRank=-=[19]-=-. Given a query,sthe retrieved results are ranked based on both page quality andsquery-page similarity.sRecently, with the rise of Web 2.0 technologies, web users withsdifferent backgrounds are creati... |
59 |
Explaining and showing broad and narrow folksonomies.
- Wal, T
- 2005
(Show Context)
Citation Context ..., a combination of “folk” and “taxonomy”, wassfirst proposed by T. V. Wal in a mailing list [12]. Folksonomyswas further divided into the narrow (e.g. flickr4) and the broads(Delicious) folksonomy in =-=[33]-=-. It provides user-created metadatasrather than the professional created and author created metadatas[2]. In [24], P. Merholz argued that a folksonomy could be quitesuseful in that it revealed the dig... |
55 | Retrieving with good sense, In:
- Sanderson
- 2000
(Show Context)
Citation Context ...er improving the performancesof SSR. However, the ambiguity problem does not affect thessearch a lot since this problem can be lightened by query wordscollocation and word senses’ skewed distribution =-=[20]-=-.s5.3 Annotation SpammingsInitially, there are few ads or spams in social annotations.sHowever, as social annotation becomes more and more popular,sthe amount of spam could drastically increase in the... |
45 | A uniform approach to accelerated pagerank computation.
- McSherry
- 2005
(Show Context)
Citation Context ...es are veryssparse in our data set, the actual time complexity is far lower.sHowever, in Web environment, the size of data are increasing at asfast speed, and some acceleration to the algorithm (like =-=[7]-=- forsPageRank) should be developed.s3.3.2 Convergence of SPR AlgorithmsHere, we give a brief proof of the convergence of the SPRsalgorithm. It can be derived from the algorithm that:s0 1 1 )()( PMMPMM... |
29 |
Using annotations in enterprise search.
- Dmitriev
- 2006
(Show Context)
Citation Context ... emergent useful information that can be used in various ways.sSome work has been done on exploring the social annotations forsfolksonomy [2], visualization [18], semantic web [36], enterprisessearch =-=[23]-=- etc. However, to the best of our knowledge, littleswork has been done on integrating this valuable information intosweb search. How to utilize the annotations effectively to improvesweb search become... |
20 |
Folksonomies: power to the people. Paper presented at the ISKO Italy-UniMIB meeting. Available at: http://www.iskoi.org/doc/folksonomies.htm
- Quintarelli
- 2005
(Show Context)
Citation Context ...ects. Hotho et al.sproposed Adapted PageRank and FolkRank to find communitiesswithin the folksonomy but have not applied them to web searchs[1]. A general introduction of folksonomy could be found in =-=[6]-=-sby E. Quintarelli.sM. Dubinko et al. considered the problem of visualizing thesevolution of tags [18]. They presented a new approach based on ascharacterization of the most interesting tags associate... |
18 | Title extraction from bodies of html documents and its application to web page retrieval,”
- Hu, Xin, et al.
- 2005
(Show Context)
Citation Context ... of them contribute from twosaspects: 1) ordering the web pages according to the querydocument similarity. State-of-the-art techniques include anchorstext generation [21, 28, 34], metadata extraction =-=[37]-=-, linksanalysis [34], and search log mining [10]; 2) ordering the webspages according to their qualities. It is also known as queryindependent ranking, or static ranking. For a long time, the staticsr... |
13 |
Metadata for the masses
- Merholz
- 2004
(Show Context)
Citation Context ...further divided into the narrow (e.g. flickr4) and the broads(Delicious) folksonomy in [33]. It provides user-created metadatasrather than the professional created and author created metadatas[2]. In =-=[24]-=-, P. Merholz argued that a folksonomy could be quitesuseful in that it revealed the digital equivalent of “desire lines”.sDesire lines were the foot-worn paths that sometimes appeared insa landscape o... |
11 |
Atomiq: Folksonomy: social classification. http://atomiq.org/archives/2004/08/folksonomy_social_classi fication.html
- Smith
(Show Context)
Citation Context ...4], visualization [18], emergent semantics [25], semantic webs[36], enterprise search [23] etc.s“Folksonomy”, a combination of “folk” and “taxonomy”, wassfirst proposed by T. V. Wal in a mailing list =-=[12]-=-. Folksonomyswas further divided into the narrow (e.g. flickr4) and the broads(Delicious) folksonomy in [33]. It provides user-created metadatasrather than the professional created and author created ... |
3 |
Okapi at TREC. In:Text REtrieval Conference
- Robertson, Walker, et al.
- 1992
(Show Context)
Citation Context ...ocuments. The average length of automatic queriessis 7.195.s4.4.2 System SetupsIn our experiment, the “DocSimilarity” is taken as the baseline.sThis similarity is calculated based on the BM25 formula =-=[29]-=-,swhose term frequency component is implemented as follows:s),()*)1((* ),(*),( dtfavgdoclendoclenbbk dtfkdtTF ++−= , (8) where f(t,d) means the term count of t in document d. In thesexperiment, k and ... |
3 |
Retrieving Web Pages using
- Westerveld, Kraaij, et al.
- 2002
(Show Context)
Citation Context ...ngsthe quality of web search. Most of them contribute from twosaspects: 1) ordering the web pages according to the querydocument similarity. State-of-the-art techniques include anchorstext generation =-=[21, 28, 34]-=-, metadata extraction [37], linksanalysis [34], and search log mining [10]; 2) ordering the webspages according to their qualities. It is also known as queryindependent ranking, or static ranking. For... |