An approach to personalized search within distributed repository of visual data

COMPUTER SYSTEMS AND NETWORKS VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE 312 © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING AN APPROACH TO PERSONALIZED SEARCH WITHIN DISTRIBUTED REPOSITORY OF VISUAL DATA And zej GLOWACZ1, Tomasz Ma cin ORZECHOWSKI2, And zej DZIECH1 1Depa men o Telecommunica ions, Facul y o Compu e Science, Elec onics and Telecommunica ions, AGH Uni e si y o Science and Technology, Al. Mickiewicza 30, 30 059 K akow, Poland 2Academic Compu e Cen e CYFRONET, AGH Uni e si y o Science and Technology, Al. Mickiewicza 30, 30 059 K akow, Poland [email p o ec ed].pl, [email p o ec ed], [email p o ec ed] Abs ac . In his pape , we in oduce unc ional assump ions o he dis ibu ed eposi o y o isual da a ha a e one o key aspec s o SYNAT amewo k o digi al lib a ies. The p oposed sys em will p o ide inno a i e ools and ex end capabili ies o cu en eposi o ies. Wi hin his ask, he pe sonalized sea ch sys em is discussed. Especially h ee di e en app oaches we e p oposed o his sys em, such as: CF, CBF and DF. Idea behind pe sonaliza ion is ha he e is cu en ly lack o such unc ionali y whe eas end use s would bene i om i . Keywo ds Digi al lib a ies, dis ibu ed eposi o y, OAI-PMH, pe sonalized sea ch. 1. In oduc ion The main goal o SYNAT p ojec [1] is he es ablishmen o he uni e sal, open, hos ing and communica ion, eposi o y pla o m o ne wo k esou ces o knowledge o be used by science, educa ion and open knowledge socie y. De eloped sys em is in ended o become wides publicly a ailable eposi o y se ice in Poland in e linking a ious digi al lib a ies a ailable and new da a se s as well. Domains o use include bu a e no limi ed o: science, echnics, educa ion, his o y and cul u e. P ojec conso ium consis s o 15 ins i u ions; among hem op polish uni e si ies and na ional lib a ies. One o key p ojec objec i es is he de elopmen o dis ibu ed eposi o y o isual da a ha will p o ide inno a i e unc ions o he sys em. This ask is conduc ed wi hin ou esea ch a eas. Fi s one is ela ed o de elopmen o dis ibu ed me ada a eposi o ies and me hods o ad anced sea ch in he ede a ion o eposi o ies o isual da a. Access in e aces will be based on common in e aces using SQI and OAI. In e linking a ious da abases u he mo e equi es uni ica ion o me ada a desc ip ion o s o ed mul imedia objec s. In addi ion, managemen o use p o iles will enable complex sea ch unc ionali ies. Second esea ch goal is he c ea ion o open da abase sys em. I will ake ad an age o on ology o design o da abase s uc u e and allow o con enien in eg a ion wi h con en -based mul imedia indexing ools. Inno a i e elemen s include de ec ion and p e en ion o dis ibu ed a acks agains da abase sys em. Ano he wo objec i es a e s ic ly ela ed o mul imedia con en and comp ised o digi al wa e ma king echnologies and me hods o high-quali y con en dis ibu ion. Conside ed digi al wa e ma king is a se o sui able echniques ha ope a e di ec ly in mul imedia con en . Thus, use ul unc ions ela ed o da a in eg i y and me ada a e i ica ion; iden i ica ion o dis ibu ion channel; o p o ec ion o sensi i e da a can be applied in eposi o y. New wa e ma king echniques a e also unde esea ch in his ask. On he o he hand, he con en dis ibu ion sys em is ocused on p o iding o end use he mul imedia da a in equi ed quali y. I will ealize unc ions o adap a ion and anscoding o isual da a, and agmen a y objec deli e y (e.g. deli e y o pa o scanned manusc ip o audio s eam om ideo eco ding). Me ada a sys em will s o e addi ional in o ma ion abou con en quali y. Func ional a chi ec u e o he eposi o y and modules a e p esen ed in Fig. 1. The emainde o he pape is o ganized as ollows. Sec ion 2 p o ides o e iew o he cu en dis ibu ion o digi al con en and exis ing sys ems o managemen o digi al con en . Sec ion 3 in oduces pe sonalized sea ch unc ionali y wi hin digi al eposi o y. Assump ions o in o ma ion il e ing, uni ied me ada a, a chi ec u e and communica ions a e p esen ed. Sec ion 4 concludes he wo k. COMPUTER SYSTEMS AND NETWORKS VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 313 Func ional F amewo k Da abase Sys em P esen a ion Laye Con en Dis ibu ion On ology o Design o Da abase Da a and Me ada a Ve i ica ion Pe sonaliza ion Mul imedia Con en Da a Video Quali y Assessmen Digi al Wa e ma king Use P o iles Me ada a and Con en -Based Mul imedia Indexing Tools Mul imedia T anscoding Fede a ed Sea ch and Da a Mining Fig. 1: Func ional a chi ec u e o he dis ibu ed eposi o y o isual da a. The emainde o he pape is o ganized as ollows. Sec ion 2 p o ides o e iew o he cu en dis ibu ion o digi al con en and exis ing sys ems o managemen o digi al con en . Sec ion 3 in oduces pe sonalized sea ch unc ionali y wi hin digi al eposi o y. Assump ions o in o ma ion il e ing, uni ied me ada a, a chi ec u e and communica ions a e p esen ed. Sec ion 4 concludes he wo k. 2. Rela ed Wo k Al hough he e a e many sea ch sys ems o digi al lib a ies, he pe sonalized sea ch app oach was no so a aken in o accoun . This was a esul o a simple and co ec assump ion ha i he cu en ca alogue sys ems we e so a e i ied du ing yea s, he digi iza ion could be he only a ans e o he exis ing model in o he on-line sys em. The de elopmen o IT echnology leads o inc easing g ow h o in e es in elec onic ma e ials. I is wo h quo ing a ollowing ac : acco ding o da a e ealed by he Amazon, as o 1 Ap il 2011, a shop sells an a e age o 105 e-books o 100 pape books [2]. The calcula ions ake in o accoun also he pape books ha ha e no elec onic equi alen , and i was no aken in o accoun he dis ibu ion o ee e-books. This ac also su p ised he p esiden o Amazon. Je Bezos said, ha he hoped ha such a s a e would happen some when, bu no one expec ed ha i had happened his ime. Especially, ha books a e sold o 15 yea s, whils e- books only om less han 4 yea s [2], [3]. Al hough he da a e e s o he U.S. Amazon, bu e-books a e also popula in he Eu opean b anches o he s o e. Go don Willoughby, he Eu opean di ec o o Kindle, said, ha e- book sales by he B i ish Amazon.co.uk we e unexpec ed. Fo example, in he UK, whe e e-books ha e been sold o only 9 mon hs and ha d-books o 13 yea s, now hey sell 242 e-books o e e y 100 ha dbacks [2], [3]. Two di e en s eps in he p ocess o digi iza ion o ypical lib a ies could be dis inguished:  digi aliza ion o lib a ian ca alogues,  digi iza ion o documen s and o he lib a ian esou ces (pain ings, ilms, e c.). Digi aliza ion o esou ces s o ed in lib a ies allows o implemen a ion o addi ional me hods o indexing con en and hus i o gene a ion o he app op ia e indexes ega dless o p e iously used me hods o ca aloguing. In o he wo ds, he me ada a ep esen ing lib a ian esou ces can now be en iched wi h he esul s o he p ocess o lib a ies esou ces indexing. A he same ime, he inc eased in e es in dis ibu ion o ideo con en can be obse ed. Al hough he e es ial and sa elli e digi al TV p o ide s ha e o e ed VoD se ices o las yea s, only now, hanks o he inc ease o bandwid h and he numbe o b oadband In e ne use s, he implemen a ion o such se ices is obse ed. The desc ip ion o he mos popula sys ems o managemen o digi al lib a ies is summa ized in Tab. 1. The echnology e olu ion in mobile elephony bo h in e ms o ansmission speed and equipmen enables cu en sma phones’ use s no only o make ideo calls, bu also soon o wa ch TV in he DVB-H s anda d. Exis ing echnologies o image and sound analysis allow o indexing o he ideo and audio sequences. They a e used o desc ibe he con en , and e en he cha ac e o he gi en sequence. Rega ding o he desc ip ion o he na u e o con en , i is wo h o men ion, ha he web po als, which o e he music ailo ed o he lis ene s' mood, exis nowadays. Pe sonaliza ion was no a key elemen in any sys ems managing digi al lib a ies ha we e shown in Tab. 1, so he p oposal p esen ed in his pape has an inno a i e cha ac e . Despi e signi ican di e ences in he cha ac e is ics be ween dis ibu ed e-lea ning sys ems, and digi al lib a ies, adap a ion o such a solu ion o he equi emen s o eposi o ies seems o be desi able and will o e a eal pe sonalized sea ch o digi al lib a ies’ use s. Tab.1: Summa y o selec ed sys ems used o managemen o digi al con en . Sys em’ s name Desc ip ion Suppo ed echnologies License ype CONTENTdm The sys em o managemen o digi al collec ions. The so wa e can handle he s o age o di e en digi al objec including: local his o y a chi es, newspape s, books, maps, slide lib a ies o audio/ ideo. I suppo s also Op ical Cha ac e Recogni ion (OCR) on ex documen s o enable ull ex sea ching [4]. Unicode, Z39.50, Quali ied Dublin Co e, VRA, XML, JPEG2000, OAI-PMH. Comme cial so wa e COMPUTER SYSTEMS AND NETWORKS VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE 314 © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING dLib a The sys em o building digi al lib a ies, de eloped by he Poznan Supe compu ing Cen e - Ne wo k (PSNC). The in en ion was o implemen he amewo k o digi al lib a ies in PIONIER ne wo k and o es ablished he Polish Fede a ion o Digi al Lib a ies. This sys em is well known and commonly used in Poland [7]. MARC, Dublin Co e, OAI-PMH Comme cial so wa e DSpace The package p o ides he ools o managemen o digi al asse s and i enables open sha ing o con en ha spans o ganiza ions. I suppo s a wide a ie y o da a, including books, heses, 3D digi al scans o objec s, pho og aphs, ilm, ideo, esea ch da a se s and o he o ms o con en [5]. METS, OAI-PMH Open sou ce so wa e unde BSD License ExLib is DigiTool The so wa e is used o managing and showcasing o Digi al Collec ions and Ins i u ional Reposi o ies. I allows o : ca aloguing, managing, sha ing, sea ching, and e ie al o digi al collec ions [6]. MARC 21, Z39.50, Quali ied Dublin Co e, METS, Z39.87-2002, OAI-PMH Comme cial so wa e FEDORA Flexible Ex ensible Digi al Objec Reposi o y A chi ec u e ha p o ides a managemen laye o digi al objec s based on con en models, which ep esen da a objec s, o collec ions o da a objec s (e.g., digi al images, XML iles, me ada a) [8]. REST/SOAP, SPARQL OAI-PMH Open sou ce so wa e unde Apache License 2.0 G eens one Digi al Lib a y I is a sui e o so wa e o building and dis ibu ing digi al lib a y collec ions. I enables he cons uc ion and p esen a ion o in o ma ion collec ions wi h e ec i e ull- ex sea ching and me ada a-based b owsing acili ies. The sys em is ex ensible: so wa e "plugins" accommoda e di e en documen and me ada a ypes [9]. Z39.50 Open sou ce so wa e unde Apache License 2.0 INVENIO I is a so wa e sui e enabling o un digi al lib a y o documen eposi o y on he web. The echnology o e ed documen inges ion h ough classi ica ion, indexing, and dissemina ion [10]. MARC21, OAI-PMH 2.0 Open sou ce so wa e unde GPL License 2.0 SimpleDL I is a digi al collec ion managemen so wa e ha suppo s di e en ype and o ma o handling documen s, such as: images, ideos, audio iles [11]. OAI-PMH Comme cial so wa e 3. Pe sonalized Sea ch Pe sonalized app oach assumed he in oduc ion o di e en me hods o ob ain, du ing he sea ch, esul s ailo ed o he needs o speci ic end-use . The e a e di e en app oaches in he implemen a ion o pe sonalized sea ch, bu gene ally he app oaches based on il a ion esul s could be dis inguished. Fil a ion is o ejec o change he o al con o mi y alue o he espec i e objec s ( om he esul se ) conside ing bo h con o mi y o he gi en que y and gi en use 's p e e ence. 3.1. In o ma ion Fil e ing and Uni ied Me ada a I is in ended o es ablish modula ized amewo k o he sys em (Fig. 2) ha could suppo all men ioned abo e app oaches. The de ails o hese well-known solu ions a e deeply analysed and desc ibed in many scien i ic a icles, compa e: [15], [16], [17]. The mos equen ly used il e ing me hods include [15]:  CF (Collabo a i e Fil e ing),  CBF (Con en -Based Fil e ing),  DF (Demog aphic Fil e ing). I is in ended o es ablish modula ized amewo k o he sys em (Fig. 2) ha could suppo all men ioned abo e app oaches. The de ails o hese well-known solu ions a e deeply analysed and desc ibed in many scien i ic a icles, compa e: [15], [16], [17]. The e a e wo he mos popula s anda d desc ibing me ada a o digi al esou ces:  desc ibing e-lea ning objec s: IEEE LOM [12],  desc ibing digi al lib a ies’ objec s: MARC [13]. The BSS (B oke Se ice Sys em), (Fig. 2) is esponsible o pa sing all hese ype o me ada a s anda ds. The de ails o componen s o he p oposed amewo k a e p esen ed in he subsec ion: A chi ec u e. Ano he me ada a s anda d, ha is in ended o be suppo ed by he ASMS (Ad ance Sea ch Managemen Sys em), (Fig. 2), is A en ion Me ada a [14]. The main pu pose o his s anda d is o exchange he in o ma ion desc ibing use s’ a en ions o sea ch esul s. 3.2. A chi ec u e The amewo k o he p oposed sys em is shown in Fig. 2. We can dis inguish ollowing pa s o he amewo k:  Sea ch Sys em – he main pa ha is esponsible o ga he ing que ies om end-use s, ans e ing hem o ano he module such as BSS o i s ex ension: Ad ance Sea ch Sys em. Then i is also esponsible o p esen ing he esul s o end-use s. COMPUTER SYSTEMS AND NETWORKS VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 315  B oke Se ice Sys em – is esponsible o o ganising he da a o que y o be sen ia he b oke laye o eposi o ies. The BSS suppo s di e en s anda ds desc ibing me ada a (including: IEEE LOM and MARC 21), di e en que y languages (including: CQL and S2QL), as well as di e en se ices (SOAP, REST) and in e aces: SQI, SRU and SRW.  B oke – is esponsible o connec ion o eposi o ies and o ganize he ede a ed sea ch. Ad ance Sea ch Sys em – is he co e pa o pe sonaliza ion. The posi ioning o ound esul s is he co e aspec o he sea ch done by gi en end-use , because i can ul il equi emen s o pe sonalized sea ch. This sys em consis s o some sub-modules including: module esponsible o s o ing end-use p o iles, module ha collec s end-use s beha iou da a, module ha collec he end-use s’ a en ion o ound esul s. All hese modules lead o o e he CF o ien ed pe sonaliza ion, whe e end- use s a e di ided in o clus e s acco ding o he simila i y, ei he o hei beha iou o o hei p o iles. Fig. 2: The model o he amewo k o he pe sonalized sea ch sys em. The e a e also some modules used by he adminis a o o manage he b oke , use s accoun s and ad ance sea ch sys ems (Fig. 2). They a e called: B oke Managemen Sys ems, Use Managemen Sys em and Ad ance Sea ch Managemen Sys em. 3.3. Communica ions The Communica ion done by b oke (Fig. 2) includes: communica ion wi h eposi o ies ha suppo SOAP Se ices including implemen a ion o SQI o e SOAP (S2QL suppo ) and he successo s o Z39.50 s anda d (suppo ing CQL): SRU (Sea ch/Re ie e ia URL) and he SRW (Sea ch/Re ie e Web se ice). Mo eo e , many o eposi o ies suppo OAI- PMH, so i is also planned o ex end he module o Ad anced Sea ch by in oduc ion ollowing sub-modules: collec ing module and indexing me ada a module. The i s one will be esponsible o he OAI-PMH communica ion wi h eposi o ies, da a ha es ing and s o ing, whils he second one will be esponsible o he me ada a indexing (including bo h IEEE LOM and MARC21 s anda ds) and o e ing he ecommenda ion u ilizing Con en -Based Fil e ing. 4. Conclusion The p oposed solu ion consis s on in oduc ion he pe sonaliza ion sea ch o digi al lib a ies’ use s. The pe sonalized sea ch is no commonly o e ed by any o exis ing so wa e o managing digi al lib a ies. Sys em will u ilize he collabo a i e il e ing by o ganize end- use s in o wo di e en kinds o clus e s on he base o wo di e en p o iles:  Dynamic: c ea ed acco ding o end-use s’ beha io ,  S a ic: connec ed o end-use s explici ly de ined p o iles. The p oposed sys em will no only o e pe sonalized sea ch o digi al lib a ies, bu also o he con en s o ed in e-lea ning eposi o ies. I is also planned o o e pe sonaliza ion ex ension by implemen ing bo h Collabo a i e Fil e ing and Con en -Based Fil e ing and inally ob ain he bes e ec o Pe sonalized Sea ch. Acknowledgemen s Wo k inanced by The Na ional Cen e o Resea ch and De elopmen (NCBiR) wi hin SYNAT p ojec no. SP/I/1/77065/10. Re e ences [1] SYNAT P ojec . Syna : Sys em nauki i echniki [online]. 2012. A ailable a : h p://www.syna .pl/. [2] GABATT, Adam. Amazon and Wa e s ones epo downloads eclipsing p in ed book sales: Success o Kindle elec onic eade p omp s apid ise o ebooks, wi h UK en husiasm ou s ipping US. In: The Gua dian [online]. 2011. A ailable a : h p://www.gua dian.co.uk/books/2011/may/19/amazon- wa e s ones-ebook-sales. [3] PRIGG, Ma k. A e only nine mon hs o Kindle, ebooks ou sell ha dbacks a Amazon. In: The Gua dian [online]. 2011. A ailable a : www. hisislondon.co.uk/s anda d/a icle- 23952130-a e -only-nine-mon hs-o -kindle-ebooks-ou sell- ha dbacks-a -amazon. COMPUTER SYSTEMS AND NETWORKS VOLUME: 10 | NUMBER: 4 | 2012 | SPECIAL ISSUE 316 © 2012 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING [4] CONTENTdm: Digi al Collec ion Managemen So wa e. OCLC: The wo ld's lib a ies [online]. 2012. A ailable a : h p://www.con en dm.o g/. [5] DSPACE. DSpace [online]. 2012 [ci . 2012-11-01]. A ailable a : h p://www.dspace.o g/. [6] ExLib is: DigiTool. ExLib is [online]. 2012. A ailable a : www.exlib isg oup.com/ca ego y/DigiToolO e iew. [7] DLib a: Digi al Lib a y F amewo k. DLib a [online]. 2012. A ailable a : dlib a.psnc.pl. [8] FEDORA: Flexible Ex ensible Digi al Objec Reposi o y A chi ec u e. Fedo a Commons [online]. 2012 [ci . 2012-11- 01]. A ailable a : h p:// edo a-commons.o g. [9] G eens one: Digi al Lib a y So wa e. G eens one [online]. 2011. A ailable a : www.g eens one.o g. [10] In enio. In enio [online]. 2012. A ailable a : in enio- so wa e.o g. [11] SimpleDL: Digi al Lib a ies Simpli ied. SimpleDL [online]. 2012. A ailable a : simpledl.com. [12] IEEE P1484.12.3. D a S anda d o Lea ning Technology: Ex ensible Ma kup Language (XML) Schema De ini ion Language Binding o Lea ning Objec Me ada a. New Yo k: IEEE, 2005. A ailable a : l sc.ieee.o g/wg12/ iles/IEEE_1484_12_03_d8_submi ed.pd . [13] Ma c S anda ds. Ma c 21 [online]. 2012 [ci . 2012-11-01]. A ailable a : h p://www.loc.go /ma c/. [14] OCHOA, X. and E. DUVAL. Use o con ex ualized a en ion me ada a o anking and ecommending lea ning objec s. In: P oceeding o he 1s In e na ional wo kshop on Con ex ualized A en ion Me ada a (CAMA’2006). A ling on: ACM, 2006, pp. 9-16. ISBN 1-59593-524-X. DOI: 10.1145/1183604.1183608. [15] MONTANER, M., B. LOPEZ and J. L. DE LA ROSA. A Taxonomy o Recommende Agen s on he In e ne . A i icial In elligence Re iew. 2003, ol. 19, iss. 4, pp. 285-330. ISSN 1573-7462. DOI: 10.1023/A:1022850703159. [16] ADOMAVICIUS, G. and A. TUZHILIN. Towa d he nex gene a ion o ecommende sys ems: a su ey o he s a e-o - he-a and possible ex ensions. The IEEE ansac ions on knowledge and da a enginee ing. 2005, ol. 17, iss. 6, pp. 734- 749. ISSN 1041-4347. DOI: 10.1109/TKDE.2005.99. [17] PAZZANI, M. A F amewo k o Collabo a i e, Con en -Based, and Demog aphic Fil e ing. A i icial In elligence Re iew. 1999, ol. 13, iss. 5-6, pp. 393–408. ISSN 1573-7462. DOI: 10.1023/A:1006544522159. Abou Au ho s And zej GLOWACZ ecei ed his Ph.D. in Telecommunica ions om he AGH Uni e si y o Science and Technology in 2007. He is cu en ly an Assis an P o esso a he AGH Uni e si y. And zej Glowacz has been wo king on nume ous comme cial p ojec s, and EU esea ch p ojec s including INSIGMA (as Depu y P ojec Coo dina o ), INDECT, GAMA, OASIS A chi e, CARMEN, DAIDALOS, DAIDALOS 2, Eu oNGI, and Eu oFGI. His main p o essional a eas o in e es a e in elligen in o ma ion sys ems, mul imedia sys ems, pa e n ecogni ion, wi eless QoS, mode n anspo p o ocols, and ad anced sys ems p og amming. He is he au ho o o e eigh y scien i ic pape s and echnical epo s; he also se es as a e iewe o in e na ional jou nals and con e ences. Tomasz Ma cin ORZECHOWSKI ecei ed his M.Sc. in Compu e Science in 1999 and Ph.D. in Telecommunica ions in 2010 om he AGH Uni e si y o Science and Technology. He is cu en ly an Assis an P o esso a he AGH-UST. His esea ch in e es s include: dis ibu ed p og amming, mobile agen echnology, da abases, ecommende sys ems and in o ma ion e ie al. He is he au ho o nume ous publica ions. He has been a membe o he IEEE Communica ions and Compu e Socie ies o o e han 10 yea s. He is he membe o se e al con e ence commi ees. And zej DZIECH g adua ed om he Lening ad Elec o echnical Ins i u e a he Facul y o Au oma ion and Compu e Science, specializing in au oma ion. in 1970 He ecei ed his Ph.D. in Telecommunica ions om he same ins i u e in 1973. He ecei ed he i le o p o esso in 1986. In 2002-2005 he was he Chai man o he Commi ee o he Sec ion o Telecommunica ions Resea ch in Poland. Cu en ly, he is he expe o Eu opean Union, e alua ing scien i ic p ojec s in he UE 7 h F amewo k P og amme. He is also he main coo dina o o EU FP7 INDECT P ojec , i led: “In elligen In o ma ion Sys em Suppo ing Obse a ion, Sea ching and De ec ion o Secu i y Ci izens in U ban En i onmen ”. He also led EU p ojec : CALIBRATE, FP6-IST - S eng hening he In eg a ion o he ICT esea ch e o in an Enla ged Eu ope.