Update-Aware Accurate XML Element Retrieval

Size: px
Start display at page:

Download "Update-Aware Accurate XML Element Retrieval"

Transcription

1 Updat-Awar Accurat XML Elmnt Rtrival Atsushi Kyaki, Jun Miyazaki Graduat School of Information Scinc Nara Institut of Scinc and Tchnology Takayama, Ikoma Nara , Japan {atsushi-k, Knji Hatano Faculty of Cultur and Information Scinc Doshisha Univrsity 1-3 Tatara-Miyakodani Kyotanab Kyoto , Japan Goshiro Yamamoto, Takafumi Taktomi, Hirokazu Kato Graduat School of Information Scinc Nara Institut of Scinc and Tchnology Takayama, Ikoma Nara , Japan {goshiro, takafumi-t, Abstract In this papr, w propos a mthod for accuratly rtriving XML lmnts considring documnt updats. If documnt updats ar not handld in a sarch systm, usrs cannot obtain appropriat sarch rsults, which rducs th usfulnss of th sarch systm. W apply an incrmntal approach to updat an indx bcaus an rbuild-from-scratch approach taks longr tim. In addition, global wights, i.. th statistics computd with all documnts in th sarch systm, may not b accurat whn a fw numbr of documnts is indxd or whn global wights chang drastically. To solv ths problms, w propos to xtnd a function of incrmntal updats of indics to gnral XML lmnt rtrival systms, with filtrs to rduc th updat cost by liminating unimportant lmnts and trms. Morovr, w apply a mthod for intgrating path xprssion which stimats accurat global wights in trm calculation. Exprimntal rsults showd that our proposd mthod can b up to 25% fastr to updat indics than th simpl incrmntal updats and can improv th sarch accuracy by 4% with documnt st of static statistics. Th proposd mthod can also sarch accuratly, vn undr continuous changs in th statistics of th documnts. I. INTRODUCTION An information unit in XML lmnt rtrival is not a documnt but an lmnt of XML documnts. XML lmnt rtrival systms prsnt lmnts that contain dscriptions satisfying th information nds of usrs who thus do not hav to spnd tim sking rlvant dscriptions in ach documnt. Although xisting studis of XML lmnt rtrival hav attaind both ffctivnss and fficincy in qury procssing [20], [22], [13], [5], ths studis hav not considrd documnt updats. Wb documnts ar frquntly updatd; i.. insrtd, dltd, or modifid. In particular, Wikipdia articls ar updatd 4000 to 8000 tims pr hour 1. Information rtrival systms ar xpctd to prsnt sarch rsults basd on th latst contnt on th Wb, spcially as nw topics ar addd to documnts. Without handling updats, a sarch systm cannot find nwly insrtd documnts, and it ranks documnts basd on obsolt information, which rducs th ffctivnss. Thus, w add a function for handling documnt updats to th xisting tchniqus for XML lmnt rtrival. Thr ar two goals for this study as follows: 1) rflcting documnt udats to a sarch systm as soon as possibl, and 1 2) computing trm wights accuratly by stimating accurat global wights. To achiv th first goal, two rquirmnts ar ndd as follows: building an indx which can b updatd incrmntally according to documnt updats, and highly fficint indxing. Th mainstram approach for updating an indx is to construct a nw indx priodically from scratch whil discarding th xisting on. It may tak a long tim to rtriv updatd documnts if constructing a nw indx is costly. Incrmntal updats ar rquird to shortn this dlay. Our proposd systm has a function of incrmntal updats of th indx. W bliv that this is th first study focusd on fast incrmntal updats of indics in ffctiv and fficint XML lmnt rtrival systms. Although Googl supports fast incrmntal updats with ffctiv and fficint qury procssing, its approach diffrs from ours. Googl analyss th link information of Wb pags to find important pags, whras our study utilizs txt information. W can apply our approach to othr structurd documnts apart from th Wb, vn if ths do not hav link information. Th proposd systm can insrt, dlt, and modify th indx by xpanding xisting an indx structur. Th updat fficincy of th indx is low bcaus a numbr of updat targts is tratd as updat targts in XML lmnt rtrival compard with documnt rtrival. Thus, w propos two kinds of filtrs for rducing updat cost by liminating unnccsary updat targts. Concrnd with th scond goal, trm wights cannot b calculatd accuratly in som situations such as: whn a sarch systm dos not stor sufficint amount of documnts, and whn nw topic mrg drastically. This is bcaus som kinds of statistics usd in calculating a trm wight ar global wights that ar aggrgat statistics drivd from all documnts in a documnt st. Thus, global wights ar difficult to calculat in th som situations mntiond abov. W nd a mthod to approximat accurat global wights with an insufficint numbr of documnts.

2 DID:1 <articl> <p>bill Gats is </p> <body> <sc>early lif </sc> <sc>windows </sc> <sc>books </sc> </body> </articl> Fig. 1. XML documnt DID:2 <articl> <sc>stv Jobs </sc> <body> <h2>businss lif </h2> <sc>appl computr </sc> </body> </articl> DID: 1, EID: 1 PE: /articl Bill Gats is Early lif Windows Books DID: 1, EID: 3 PE: /articl/body Early lif Windows Books DID: 1, EID: 2 PE: /articl/p Bill Gats is DID: 1, EID: 4 PE: /articl/body/sc Early lif DID: 1, EID: 5 PE: /articl/body/sc Windows DID:1, EID: 6 PE: /articl/body/sc Books DID:2, EID: 1 PE: /articl Stv Jobs Businss lif Appl computr DID:2, EID: 3 PE: /articl/body Businss lif Appl computr DID:2, EID: 2 PE: /articl/sc Stv Jobs DID:2, EID: 4 PE: /articl/body/h2 Businss lif DID:2, EID: 5 PE: /articl/body/sc Appl computr DID:1 DID:2 Fig. 3. XML lmnt articl EID:1 articl EID:1 p EID:2 sc EID:2 body EID:3 body EID:3 Bill Gats is... Stv Jobs... sc EID:4 sc EID:5 sc EID:6 h2 EID:4 sc EID:5 Early lif... Windows... Books... Businss lif... Appl computr... Fig. 2. XML tr Authors add structurs to a documnt:.g. chaptrs, sctions, and paragraphs. W utiliz ths structurs to idntify th bst matrial for satisfying th information nds for usrs. Som structurs ar maninglss, so lmnts dfind by thos structurs ar inappropriat as sarch rsults, and som xisting studis [4], [3] includ attmpts to liminat ths. Suppos that a usr sks information from Documnt 1 about Early lif..., Windows..., and Books. XML lmnt rtrival systms try to prsnt th usr Elmnt 3 of Documnt 1, sbcaus that lmnt contains all of th information that th usr nds and no xtra information. W valuatd th ffctivnss and fficincy of our approachs through xprimnts with two cass: th static statistics cas in which topics rarly chang, and th dynamic cas in which nw topics ar addd frquntly. In th rmaindr of this papr, Sctions II and III dscrib th basic concpt of XML lmnt rtrival and th rlatd studis, rspctivly. Sction IV discusss fast incrmntal updats of indics. Sction V rmarks accurat XML lmnt rtrival with considring documnt updats. Finally, Sction VI discusss th xprimntal valuations. Sction VII concluds this papr. II. XML ELEMENT RETRIEVAL In this sction, w dscrib th concpts of XML lmnts and quris in XML lmnt rtrival. A. XML Elmnt W giv spcific xampls in Figurs 1 3 to dfin XML lmnts. Figur 1 illustrats XML documnts. Each documnt is assignd a documnt idntifir (DID). Figur 2 dpicts trs abstractd from Figur 1. An XML documnt can b prsntd by a tr structur, which hlps to undrstand th structur of th documnt. Each lmnt is assignd an lmnt idntifir (EID), which is assignd in documnt ordr. W can idntify an lmnt using its DID and EID. A pair of start and nd tags rprsnts an XML lmnt nod within an XML tr, and th nstd structur of XML lmnts rprsnts ancstor dscndant rlationships. Each lmnt in Figur 3 is th txt that compriss a st of txt nods within th XML tr in Figur 2. W dscrib th path xprssion (PE) of ach lmnt. B. Quris for XML Documnts Thr ar two ways for xprssing an information nd in a qury: through kywords and through documnt structur. A qury ntirly composd of qury kywords is calld a contntonly (CO) qury, whras a qury composd of pairs of qury kywords and a constraint on th documnt structur is calld a contnt-and-structur (CAS) qury. CO quris ar usd just as in traditional information rtrival for txt documnts. Usrs can submit CO quris vn if thy do not know th structurs of th documnts that ar rtrivd. In contrast, CAS quris utiliz on of th most significant faturs of structurd documnts: i.. documnt structur. With a CAS qury, a usr can obtain spcific rsults with rgard to granularity and contnt. W giv a spcific xampl of a CO qury and a CAS qury. Ths ar xprssd in th narrow xtndd XPath I (NEXI) [23] qury languag. A CO qury //*[about(., "Windows")] mans that th candidat sarch rsults ar lmnts containing Windows. Elmnts 1, 3, and 5 of Documnt 1 can b sarch rsults in Figur 3. A CAS qury: //articl[about(.,"stv")]//sc [about(., "Appl")] is mor complx. Lt us focus on th first half of th qury, //articl[about(., "Stv")], which mans that candidats for this part ar lmnts that contain Stv and whos path xprssions nd with an articl tag. Th scond half of th qury, //sc[about(., "Appl")], mans that candidat sarch rsults ar lmnts that contain Appl and whos path xprssions nd with a sc tag. Th sarch rsults ar lmnts that satisfy th lattr constraint and whos ancstor lmnts satisfy th formr constraint. Th only lmnt

3 satisfying th qury constraints is Elmnt 5 of Documnt 1 in Figur 3. III. RELATED STUDIES Hr w xplain ffctiv and fficint XML rtrival. W also discuss studis focusd on updats in sarch systms. A. Effctiv and Efficint XML Sarchs 1) Effctiv XML Sarchs: Th most important goal of XML lmnt rtrival is highly accurat sarchs. Th mainstram approach to xtracting rlvant lmnts is as follows: first, calculat a trm wight for ach lmnt by using a trmwighting schm; nxt, comput a scor for ach lmnt using ths trm wights. Trm-wighting schms for XML lmnt rtrival ar oftn drivd from studis on documnt rtrival. Both of ths ar composd of thr typs of factors: local wights that ar statistics drivd from ach documnt (lmnt); global wights that ar statistics drivd from all documnt in a documnt st; and constant valus (cofficints and paramtrs). Local wights and constant valus ar asy to calculat and rfr to bcaus local wights ar computd for a nwly insrtd lmnt. Howvr, it is difficult to calculat global wights on dmand bcaus th ntir documnt st must b scannd to comput ths. Th most significant diffrnc btwn documnt rtrival and XML lmnt rtrival is th mthod for computing global wights. Trm-wighting schms in documnt rtrival assum that vry documnt has th sam attribut and blongs to th sam class. Thus, global wights ar calculatd using all documnts. Howvr, in XML lmnt rtrival, lmnts ar assignd to classs. Global wights ar calculatd for lmnts of th sam class. Thr ar diffrnt ways to classify lmnts. On approach is to classify lmnts by path xprssion. In Figur 3, sinc Elmnts 4, 5, and 6 of Documnt 1, and Elmnt 4 of Documnt 2 all hav th sam path xprssion /articl/body/sc, th global wights ar calculatd using ths lmnts. Altrnativly, lmnts with th sam tag can b placd in th sam class. Bcaus Elmnts 4, 5, and 6 of Documnt 1 and Elmnts 2 and 4 of Documnt 2 all hav th sc tag, th global wights ar calculatd using ths lmnts, as dpictd in Figur 3. W us classification basd on path xprssion in our systm, bcaus this is rportdly mor accurat [18]. Thr ar svral kinds of trm-wighting schms for XML lmnt rtrival;.g. TF-IPF [11], BM25E [12], and th qury liklihood modl for XML lmnt rtrival [9] (QLMER). BM25E is a probabilistic modl. In a trm calculation of th classic trm-wighting schm TF-IPF, statistics on th occurrnc frquncis of trms ar utilizd. Convrsly, BM25E utilizs not only th statistics but also lmnt lngth (th numbr of trms in an lmnt). Languag modl tchniqus hav bn dvlopd in th filds of spch rcognition and machin translation. Rcntly, ths tchniqus hav bn introducd into th fild of information rtrival. In particular, th qury liklihood modl [14] is wll studid and achivs significant rsults. This modl has bn adaptd to XML lmnt rtrival [9]. In th trmwighting schm, th scor of ach lmnt is th product of th occupancy probabilitis of th qury kywords as shown in Eq. (6). This mans that non-zro valus ar computd only for th lmnts containing all th qury kywords. To avoid this, smoothing tchniqus ar oftn usd. Smoothing valus ar computd not with a documnt (lmnt) modl but with a background languag modl, which is applid for an ntir documnt st. BM25E is rgardd as a mor ffctiv trm-wighting schm than TF-IPF. Actually, most of th top-rankd sarch systms at INEX us BM25E [2]. Howvr, no xhaustiv comparison btwn BM25E and QLMER has bn xplord. W thrfor xamin th potntials of ths trm-wighting schms in this articl. 2) Efficint XML Sarchs: Although th most important rquirmnt of XML lmnt rtrival is nabling ffctiv sarchs, fast qury procssing is also rquird by systm usrs. To attain fficint XML lmnt rtrival, various approachs hav bn takn, such as: applying top-k algorithms to rturn sarch rsults quickly, and comprssing and rducing data to supprss th indx siz to minimiz th amount of data scannd in qury procssing. Many top-k sarchs hav bn proposd [6]. Thr ar two conditions for fficint qury procssing: 1) trm wights ar calculatd bfor qury procssing bgins, and 2) trms ar sortd in dscnding ordr of wight. This mans that w only nd to scan highly rankd trms in qury procssing. Not that som qury procssing mthods also utiliz an indx for a random scan, which is usd to rfr to th wight of an arbitrary trm in any lmnt. Som studis [20], [22] hav usd trm-wighting schms [12] for ffctiv sarchs. Thobald t al. proposd two typs of indics and a top-k algorithm for fficint sarchs [20]. On typ is for scoring an lmnt in qury procssing, and th othr typ is for chcking a structural constraint on a qury. Thy also proposd cost-basd qury procssing, which idntifis an ffctiv momnt to chck th structural constraints and dtrmins which qury kyword is rasonabl to procss. Trotman t al. proposd a low-cost mthod of data comprssion and slction [22]. In ths studis, th aim was to rtriv lmnts satisfying th information nd of a usr by rtriving lmnts from a fixd documnt st; i.. documnt updats wr not considrd. B. Handling Documnt Updats on th Wb Th handling of documnt updats is spcially important in Wb sarch systms bcaus documnts ar constantly insrtd, dltd, and modifid. Whn documnts ar updatd, usful sarch systms should trat thm as sarch targts immdiatly. If systms prsnt sarch rsults basd on a

4 past snapshot of th Wb, th contnt of th Wb documnts may sinc hav changd. Sarch systms should rflct th currnt stat of th Wb and handl dynamically changing Wb documnts. Rcntly, som tchniqus for handling documnt updats hav bn proposd. Chn t al. [1] tackld this challng in th fild of information xtraction. Thy rportd that a long procssing tim is rquird to apply information xtraction tchniqus to documnt collctions whn documnt updats occur. As a rsult, a dlay occurs bfor information xtractd from th updatd documnts is availabl. To shortn th dlay, thy proposd a mthod for rcycling th intrmdiat rsults of past snapshots. Numann t al. [17] also ffctivly utilizd th information of past snapshots, but with Rsourc Dscription Framwork (RDF) data. Rn t al. [19] prsrvd not only th latst graph data but also past snapshots to trac th transition of th graph. Our study is diffrnt from thirs to th xtnt that w prsnt th information along with th latst stat of th Wb. Th aformntiond studis utilizd th intrmdiat rsults of past snapshots. Hnc, w also utiliz thos or xisting indics. W incrmntally updat xisting indics whn nw documnts ar insrtd. In addition, Wb sarch systms ar xpctd to maintain high prformanc with a low updat cost. In th cas of txt rtrival, high sarch accuracy should also b maintaind. Thr has bn no adquat study focusd on incrmntal updats in XML lmnt rtrival with ffctiv and fficint qury procssing. Thrfor, this is th first study to tackl th problm. Although som rsarchs hav focusd on incrmntal updats of an invrtd indx [21], [10], [15], thy proposd indx data structurs of indics and physical storag mthods. Our study diffr from thir studis bcaus w introduc a function for incrmntal updats of indics for svral purposs in XML lmnt rtrival by proposing an fficint mthod of data managmnt. IV. FAST INCREMENTAL UPDATES OF INDICES Gnral XML lmnt rtrival systms [20] [22] hav functions for indx construction and qury procssing. Thy rbuild an indx from scratch whn documt updats occur. This mans it tals long tim to build a nw indx whn th amount of accumulatd documnts is larg. In contrast, our proposd systm has capabilitis for documnt insrtion, dltion, and modification to rflct documnt updats immdiatly. Morovr, w propos two filtrs which rduc updat cost for attaining fast incrmntal updats. A. Expansion of Existing Functions 1) Trm Wighting Schm: Th proposd systm us an arbitrary trm wighting schm. Basd on th discussion in Sction III-A1, w us ithr BM25E [12] or qury liklihood modl for XML lmnt rtrival [9] (QLMER). In th BM25E, a trm wight w bm25 (p,, t) of trm t in lmnt with path xprssion p is calculatd as follows: w bm25 (p,, t) = (1) (k 1 + 1)tf,t k 1 ((1 b) + b l log N p pf p,t avl p ) + tf,t pf p,t Trm (DID, EID, trm, trm wight, Path ID, lmnt lngth) - Tag-trm (DID, EID, tag, trm, trm wight, Path ID, lmnt lngth) - RS (DID, EID, trm, trm wight) - Path (Path ID, path xprssion) - GW-Path-trm (Path ID, trm, frquncy [, valus of back-ground languag modl for QLMER]) - GW-Path (Path ID, frquncy, total lngth) - Trm-filtr (tag, trm, thrshold valu) Fig. 4. Structur of th indics whr tf,t is th trm frquncy of trm t in lmnt, pf p,t is th lmnt frquncy of trm t in th lmnts with p, N p is th numbr of lmnts with p, l is th lngth of lmnt, and avl p is th avrag lngth of th lmnts with p. Th paramtrs k 1 and b ar st as commonly usd valus 2.5 and 0.85, rspctivly, basd on our past xprimnts. Morovr, s bm25 () is th scor of and is calculatd as follows: s bm25 () = t i T w bm25 (p,, t i ) (2) whr T is a st of qury kywords. Likwis, w show th way of trm culculation with QLMER. Though thr ar som variations of QLMER, our prvious study rports that th path xprssion-basd Dirichlt smoothing approach is th most ffctiv [9]. Lt w qlm (p,, t) b a probability that trm t is gnratd in lmnt (i.. a trm wight), and lt s qlm (p,, t) b a scor of lmnt. Ths ar calculatd as follows: w(p,, t) = ˆP ml (t M ) + µ ˆP ml (t, p M b ) l + µ (3) ˆP ml (t M ) = tf,t (4) l p ˆP ml (t, p M b ) = tf,t p l (5) s qlm () = ˆP (T M ) = t T w qlm (p,, t) (6) whr M is an lmnt modl for, M b is a background languag modl for p, µ is a givn paramtr. 2) Structurs of Indics: W show th structurs of th proposd indics in Figur 4. In many xisting studis, th trm wights stord in th indics ar calculatd bforhand, and structural constraints can b chckd with ths. Th proposd indics inhrit ths capabilitis but also contain global wights to calculat a trm wight immdiatly, which is ssntial for fast incrmntal updats. As in th rlatd studis [20], [22], th structurs of th indics ar dfind in an RDB format. Primary kys ar undrlind. Th Trm, th Tag-trm, th RS, and th Path

5 indics ar usd for fficint and ffctiv qury procssing as in othr studis. In th GW-Path-trm and th GW-Path indics, th global wights ar indxd for nabling fast trm calculation. In Eq. (1), (4), and (5), pf p,t, N p, avl p, and ˆP ml (t M b ) ar global wights. W discuss th Trm-filtr indx blow in Sction IV-C2. 3) Top-k Sarchs: Th proposd systm has a function for top-k sarchs [6] to nhanc its usability in fast qury procssing. To rturn sarch rsults, only th top k tupls ar rtrivd for ach trm (a pair of tag and trm for CAS quris). Th trm wights in th tupls ar summd for ach lmnt to calculat scors. Furthrmor, a wight in an arbitrary trm in any lmnt can b gaind with a random scan whn w nd to calculat xact scors for sarch rsults. W can attain not only fficint qury procssing but also ffctiv qury procssing with th random scan. A CO qury rtrivs th Trm indx whras a CAS qury rtrivs th Tag-trm squntially to xtract candidat sarch rsults in qury procssing. Accurat scors ar calculatd for lmnts by a random scan with th RS indx. Not that tupls in th Trm indx ar groupd by trm in dscnding ordr of trm wight, whras tupls in th Tag-trm indx ar groupd by pair of tag and trm. Whn a CAS qury contains two or mor structural constraints, th path xprssions of lmnts must b chckd to dtrmin whthr ths satisfy th qury constraints. B. Handling Documnt Updats 1) Documnt Insrtion: Whn a documnt is insrtd, th updating procss is conductd as follows: (1) xtracting lmnts from th insrtd documnt, (2) calculating trm wights for th lmnts, and (3) updating indics. First, th documnt is parsd and lmnts ar xtractd. As a rsult, lmnts 1 and 2 ar xtractd. Scond, th trm wights of t 1 and t 2 in 1, and t 1 in 2 nd to b calculatd. Trm wights ar calculatd immdiatly with th GW-Path-trm and th GW-Path indics. W only nd to stor all kinds of global wights in th indics whn using anothr trm-wighting schm rquiring othr statistics. Finally, th Tag-trm, th Trm, and th RS indics ar updatd incrmntally aftr th trm wights ar calculatd. Not that an ntir st of documnts can b updatd at onc to rduc th I/O cost. 2) Documnt Dltion: Whn a documnt is dltd, thr is a high cost to find and dlt all tupls rlatd to th documnt bcaus th tupls ar sprad across th indics. W thrfor tak anothr approach to rduc th cost of th dltion. W manag th DIDs of dltd documnts instad of dlting tupls in th indics. Thn, w simply ignor th tupls of th DIDs in qury procssing. With this approach, w can rflct th dltion immdiatly. Fig. 5. Vrsion list DID VID dltd indx DID EID VID modifid th vrsion list and qury procssing sarch rsults DID EID VID W prpar a vrsion list to manag th dltd documnts. Th list contains pairs. Each pair contains th DID of th dltd documnt and th vrsion idntifir (VID) with its valu markd as. W ovrwrit th VID as whn th DID of th dltd documnt is containd. Spcifically, Documnt 100 in Figur 5 has bn dltd bcaus th VID of Documnt 100 is. Th tupls of documnts dltd in th indics ar liminatd whn th load avrag is low. Aftr liminating th tupls, th DIDs of th documnts dltd in th vrsion list ar also liminatd. 3) Documnt Modification: Th modification procss is achivd through th dltion and insrtion procsss. In mor dtail, w dlt all tupls rlatd to th modifid documnt and insrt th latst vrsion of th documnt to handl th modification. W also utiliz th vrsion list to manag th vrsion of th documnt, bcaus thr is a high cost to dlt th tupls of a modifid documnt immdiatly. To nabl fast modification, w only targt th tupls of th latst vrsion in qury procssing. Not that th granularity of modification is documnt granularity, bcaus som problms aris with lmnt granularity. On of th problms is th siz of th vrsion list. Th ovrhad in qury procssing bcom gratr whn w manag not documnts but lmnts. Anothr problm is th difficulty in mapping old structur to nw structur whn th documnt structur changs. Ths ar th rasons that w adopt th documnt as th granul of modification. Th modification procss is conductd as follows: first, whn a modification occurs, th vrsion list is scannd to dtrmin whthr th DID of th modifid documnt is containd. If th DID xists in th vrsion list, 1 is addd to th VID; othrwis, th DID of th modifid documnt and its VID valu of 1 ar insrtd. For xampl, Documnt 101 in Figur 5 has gon through modification twic bcaus its VID is 2. Scond, th Trm, th Tag-trm, and th RS indics ar updatd in th sam mannr as for documnt insrtion. As shown in Figur 4, ach tupl contains a VID whos valu is th sam as that writtn in th vrsion list. Not that th VID of th first documnt insrtd is 0. Finally, ach tupl is chckd to dtrmin whthr th tupl is valid in qury procssing basd on th VID. Th tupl is th latst whn th VID of th tupl is th sam as that of th modifid documnt in th vrsion list. Morovr, th tupl is also th latst whn th DID of th modifid documnt is

6 not containd in th vrsion list. In contrast, th tupl is invalid whn th VID of th tupl is smallr than that of th modifid documnt in th vrsion list. W giv a spcific xampl of th validation chck in Figur 5. Th DID of th first tupl in th indx is 103, and th vrsion list dos not contain that DID. Thus, th first lin is valid. Th documnt of th scond tupl has bn dltd, bcaus th DID of this tupl is containd in th vrsion list and its VID is. Th third tupl is th latst, bcaus th VID of this tupl is th sam as th VID corrsponding in th vrsion list to th Documnt 101. Similarly, th fourth tupl is not th latst, bcaus th VID of this tupl is lss than that of th VID corrsponding to th Documnt 101. Old vrsions of tupls in th indics ar rmovd whn th load avrag is low. In this rgard, th VID of th latst vrsion of a tupl is rwrittn as 0, and th DID of th dltd documnt is rmovd from th vrsion list. C. Filtrs for Rducing Updat Cost W propos two kinds of filtrs for slcting important lmnts and trms to indx. It is obvious that w can rduc updat cost with ths filtrs. Howvr, sarch accuracy will b rducd if w rmov lmnts and trms rlvant to any qury. This would violat th first rquirmnt. To avoid a dcras in sarch accuracy, w should dcid carfully which lmnts and trms can b rmovd. 1) Elmnt Filtr: W propos an lmnt filtr to rmov unncssary lmnts. W prviously proposd a mthod to rmov lmnts that cannot b th sarch rsults of any qury [3] and a mthod to idntify th most appropriat granularity for sarch rsults [8]. Thos studis ld to th fact that modrat granuls ar th most appropriat as sarch rsults, bcaus xtrmly larg granuls (.g. whol documnts) tnd to contain irrlvant dscriptions and xtrmly small granuls cannot satisfy th information nd by thmslvs. Hrinaftr, w attmpt to rmov xtrmly small lmnts, sinc idntifying ths is asir. It is ssntial to dfin what xtrmly small lmnts ar. Many of th Wb documnts includ tabl-of-contnts or rfrnc information, which basically consists not of sntncs but of kywords. Ths dscriptions cannot satisfy an information nd dirctly, although thy can srv as navigational information. Sinc on rquirmnt for txt summarization is that information should b slf-containd [14], w rmov any lmnt that cannot b undrstood by itslf. Basd on th discussion abov, w dfin thr conditions of xtrmly small lmnts as follows: (1) lmnts containing fw trms (thrshold τ l ), (2) lmnts with dp path xprssions (thrshold τ dpth ), and (3) lmnts with rar path xprssions (thrshold τ Zipf ). Rgarding th first condition, th trms in th information othr than th body txt including tabl-of-contnts and rfrnc information, contain fw trms also in th lmnts. Actually, study [3] rports that sarch accuracy improvs whn short lmnts ar rmovd. Fig small trms 2 dp path 3 4 rarly apparing path l m n t filt r trm calculat ion - GW-Path-trm - GW-Path Th lmnt filtr and th trm filtr thrshold valu : 0.5 t 1 :1.0 t 2 : 0.3 t 3 : RS t r m filt r - Trm - Tag-trm In th scond condition, lmnts with dp path xprssions ar liminatd. Tabls or lists in HTML hav a tndncy to b nstd dply. Th valu of ach cll is nonsns without furthr information, which is th rason that w rgard ths lmnts as irrlvant. Rgarding th third condition, path xprssions that rarly appar in th documnt st cannot b calculatd accuratly, as discussd abov in Sction V-B. W thrfor us Zipf s law [16] to obtain th thrshold of mdian frquncy f, which is computd as follows: 8F f = (7) 2 whr F 1 is numbr of th path xprssions apparing only onc in th documnt st. To rtain sarch accuracy, w sk appropriat thrsholds to rmov only irrlvant lmnts. Prliminary xprimnts on th lmnt filtr ar dscribd in Sction VI-B. Figur 6 illustrats th bhavior of th lmnt filtr. Suppos that four lmnts, 1, 2, 3, and 4, ar xtractd from insrtd documnts. Elmnts 1, 2, and 4 ar liminatd by th lmnt filtr bcaus 1 is too short, th path xprssion of 2 is too dp, and th path xprssion of 4 rarly appars. As a rsult, only 3 is chosn as a targt. 2) Trm Filtr: Although thr ar many candidat sarch rsults, only a fw lmnts ar prsntd as sarch rsults. Thrfor, w suppos that sarch accuracy is not significantly affctd if indics do not contain trms with low wights. Basd on this ida, w rmov th unimportant trms with th trm filtr. Th thrsholds τ tw ar dfind as th trm wights of th nth largst trm for ach pair of tag and trm containd in th indics. Ths valus ar stord in th Trm-filtr indx so that thy can b lookd up quickly. Not that w apply th trm filtr only to th Trm and Tag-trm indics to nabl accurat calculation of th scor for lmnts with th RS indx. In addition, w do not apply th filtr whn th numbr of tupls of th pair of tag and trm is lss than n. Figur 6 shows an xampl of how th trm filtr works. Suppos that τ tw is 0.5 and thr ar thr trms to insrt into th Tag-trm, th Trm, and th RS indics. W us th singl valu of τ tw for simplicity although τ tw diffrs for ach

7 pair of tag and trm. Trms t 1 (1.0 > τ tw ) and t 3 (0.8 > τ tw ) ar succssfully indxd with th Trm, th Tag-trm, and th RS indics bcaus thy ar gratr than τ tw. In contrast, trm t 2 (0.3 < τ tw ) is only indxd with th RS indx bcaus it is lss than τ tw. V. ESTIMATING ACCURATE GLOBAL WEIGHTS A. Effcts of Incrmntal Updats W xamin th ffctivnss and fficincy of incrmntal updats of indics. 1) Tst Collction and Implmntation Sttings: In th xprimnts, w usd th INEX 2008 tst collction providd by th INEX projct 2. This tst collction consists of thr componnts: (1) th INEX documnt collction, (2) th INEX topics, and (3) th INEX rlvanc assssmnts. Th INEX documnt collction is an XML Wikipdia corpus basd on a snapshot of th English vrsion of Wikipdia. Approximatly 660,000 articls ar in this corpus. Th INEX topics includ 68 quris, of which 32 ar CO quris and 36 ar CAS quris. W usd all of ths in th xprimnts. Th INEX rlvanc assssmnts ar th valuations of th quris to masur th ffctivnss of XML lmnt rtrival systms. In this tst collction, at most 1,500 lmnts ar prsntd as sarch rsults for ach qury. In th INEX projct, th intrpolatd prcision at th rcall lvl of 1% (ip[.01]) is usd as a formal masur of accuracy. Th valuation tool also masurs th man avrag intrpolatd prcision (MAiP) as th avrag prcision at 101 rcall lvls. Th PC that w usd for th xprimnts runs Oracl Entrpris Linux 5.5. It has four Intl Xon X7560 CPUs (2.3GHz), 512GB of mmory, and a 4.5TB disk array. Th indics wr implmntd using BrklyDB in GNU C++. 2) Exprimntal Procdur: W dfin an indx bfor incrmntal updats tak plac as an initial indx. W distinguish btwn documnts usd to construct initial indics (initial documnts) and documnts usd to updat indics (updat documnts). Hr, w assum that th statistics of th documnts ar static, i.., th statistics of th initial documnts and th updat documnts ar th sam. For this purpos, w randomly sampld documnts in ordr to distinguish btwn thm. In Sction VI-D, w considr a mor complx cas in which th statistics of th documnts chang dynamically. All documnts ar procssd through th stop-word and stmming stps bfor th construction of th initial indics bgins. Th procdur is as follows: first, th initial documnts ar parsd to calculat trm wights and th initial indics ar constructd; thn, th updat documnts ar obtaind for updating indics incrmntally. All data in th GW-Path-trm and th GW-Path indics ar scannd in th main mmory during updats. Thn, th updat documnts ar parsd and th Trm, th Tag-trm, and th RS indics ar updatd incrmntally. 2 TABLE I. ACCURACIES OF TERM WEIGHTING SCHEMES BM25E QLMER (µ) ip[.01] ) Choosing a Trm-wighting Schm: W conductd a prliminary xprimnt to choos a trm-wighting schm usd in latr xprimnts. W xamind th ffctivnss of BM25E and QLMER to ascrtain which trm-wighting schm is mor accurat on. Tabl I indicats that BM25E is mor ffctiv trmwighting schm. Hnc, w usd this in th latr xprimnts. 4) Evaluation Rsults: W invstigatd sarch accuracis, updat fficincy pr documnt, and total tim of indx construction by changing th prcntag of initial documnts within th documnt st, as indicatd in Tabl II. For xampl, whn th ratio is 30%, th initial indics ar constructd using 30% of th documnts in th st, and th indics ar updatd using th rmaining 70% of th documnts. Whn th ratio of initial documnts is 100%, updats of th indics do not tak plac (no-updat). Tabl II shows that incrmntal updats rduc sarch accuracy, which dmonstrats that global wights cannot b computd accuratly using only a subst of th documnts. To mak th incrmntal updat practical, w nd to solv th problm of inaccurat global wights. Th avrag tim for incrmntal updats is 53.4 ms pr documnt whn th ratio of initial documnts is 50%, whras th tim rquird to construct indics from scratch (no-updat) is 42.1 ms pr documnt. This suggsts that th updat fficincy dcrass as th ratio of initial documnts incras. As a rsult, indxing may tak a long tim whn w updat a numbr of documnts. B. Intgrating Path Exprssion for Accurat Global Wights W attmpt to calculat global wights accuratly using a limitd numbr of documnts. Sinc ths ar calculatd within lmnts having th sam path xprssion, w cannot obtain appropriat statistics for a path xprssion apparing rarly in th documnt st. W thrfor considr a mor ffctiv approach. Spcifically, w intgrat path xprssions having a similar proprty to xpand th lmnts in th sam class. To accomplish this, w utiliz th mthod proposd in our prvious study [7] for intgrating path xprssions. This intgration mthod calculats an accurat global wight for a path xprssion of fw frquncis. Th currnt cas is similar to that in th prvious study. In both cass, th global wights of lmnts with rar path xprssions ar not calculatd accuratly. Thrfor, th intgration mthod should improv th rsults. To intgrat path xprssions, w rgard a path xprssion as an array of tags and idntify th path xprssions that ar similar to ach othr in trms of th apparanc ordr and apparanc frquncis of tags. As a rsult of th intgration, w liminat classs that do not contain nough lmnts to calculat accurat global wights.

8 TABLE II. THE RESULTS OF THE SIMPLE APPROACH ratio of th initial ip[.01] MAiP updat tim total tim of documnts (%) (ms/doc) indx construction (h) from-scratch : /articl/sc 2: /articl/sc/sc 3: /articl/sc/mp/sc 4: /articl/mp/sc 5: /articl/mp/sc/sc articl: 1, sc: 1 articl: 1, sc: 2 articl: 1, sc: 2, mp: 1 1: /articl/sc 2: /articl/sc/sc 3: /articl/sc/mp/sc 5: /articl/mp/sc/sc Fig. 7. Exampls of path xprssions articl: 1, sc: 1, mp: 1 4: /articl/mp/sc Fig. 8. articl sc articl sc mp 1: /articl/sc 2: /articl/sc/sc 3: /articl/sc/mp/sc 4: /articl/mp/sc 5: /articl/mp/sc/sc An xampl of classification in ST Fig. 9. An xampl of classification in BT /articl+/sc+ /articl+/sc+/mp+/sc+ /articl+/mp+/sc+ 1: /articl/sc 2: /articl/sc/sc 3: /articl/sc/mp/sc 4: /articl/mp/sc 5: /articl/mp/sc/sc In addition, th cost to adopt ths mthods is small, bcaus ths approachs simply calculat a frquncis and chck th ordr of tags in a path xprssion. W can ignor th harmful ffcts on updat fficincy. W now xplain thr intgration mthods: 1) st-of-tags mthod (ST), 2) bag-of-tags mthod (BT), and 3) ordr-of-tags mthod (OT). 1) St-of-Tags Mthod (ST): Tags in structurd documnts ar sparatd into two groups. On rprsnts structural classifications such as articl and sc tags. Th othr indicats smantics, idas, attributs, and spcific contnts such as prson, mp, and tabl tags. Ths two groups of tags ar supposdly indpndnt in thir apparanc. This suggsts that a combination of tags can gnrat two or mor path xprssions. It is not always appropriat that ths path xprssions ar placd into diffrnt classs. This is why w focus on rlaxing th apparanc ordr and frquncis of tags in path xprssions to intgrat similar path xprssions. Th st-of-tags (ST) mthod rlaxs both th apparanc ordr and frquncis of tags in path xprssions. Accordingly, w considr only th nams of th tags. W classify path xprssions composd of th sam tag nams as mmbrs of th sam class. Classification of th path xprssions in Figur 7 is shown in Figur 8. Th first two path xprssions ar in th sam class bcaus thy ar both composd of articl and sc tags, whil th othr thr path xprssions ar in th sam Fig. 10. An xampl of classification in OT class bcaus thy ar all composd of articl, sc, and mp tags. Th global wights of th lmnts with th first two path xprssions ar calculatd togthr, and th global wights of th lmnts with th othr path xprssions ar calculatd togthr. 2) Bag-of-Tags Mthod (BT): Th bag-of-tags (BT) mthod rlaxs only th apparanc ordr of tags in path xprssions. W do not considr th ordr of tags from th prspctiv of th bag-of-words concpt. Classification of th path xprssions in Figur 7 is shown in Figur 9. W first numrat th nams and frquncis of tags in ach path xprssion to intgrat th path xprssions classifid as mmbrs of th sam class. As a rsult, w intgrat th third and fifth path xprssions bcaus both hav on articl, two sc, and on mp tags. 3) Ordr-of-Tags Mthod (OT): Th ordr-of-tags (OT) mthod rlaxs only th apparanc frquncis of squntial tags in a path xprssion. In som path xprssions, a tag appars conscutivly two or mor tims; for xampl, col tags in tabl of HTML. In this cas, vn if th frquncis of a tag apparing conscutivly ar diffrnt, w suppos that th faturs of a path xprssion ar not much diffrnt bcaus th smantics of ach tag ar fixd. Thrfor, if conscutiv tags ar th sam, such tags can b aggrgatd. Classification of th path xprssions in Figur 7 is shown in Figur 10. Not that sc tags appar conscutivly in th scond and fifth path xprssions. Th first and scond path xprssions ar intgratd, bcaus ths hav hav on or

9 TABLE III. ACCURACIES WITH CHANGING τ l τ l ip[.01] TABLE V. EFFECTS OF THE TERM FILTER WITH CHANGING n n no filtr ip[.01] TABLE IV. DEPTH OF PES AND THE RATIO OF ELEMENTS τ dpth All Top mor articl tags followd by on or mor sc tags. Th fourth and fifth path xprssions ar also intgratd, ths hav on or mor articl tags followd by on or mor mp tags, and on or mor sc tags. VI. A. Exprimntal Dsign EXPERIMENTAL EVALUATIONS With th simpl incrmntal updat systm as th baslin, w invstigat whthr th applying two filtrs and intgrating path xprssions ar ffctiv for sarching accuratly and fficint for updating th indics or not. Th xprimntal nvironmnt is th sam as that usd in Sction V. Th proposd mthods ar valuatd using two documnt sts; on with static statistics, which mans that its topics rarly chang; and th othr with dynamic statistics, which mans that nw topics ar rgularly addd. Our proposd approachs admit som variations. Thr ar four ways of calculating global wights: th dfault mthod, which is classification basd on path xprssion; th stof-tags mthod (ST); th bag-of-tags mthod (BT), and th ordr-of-tags mthod (OT). Thr ar thr paramtrs in th lmnt filtr: th lmnt lngth thrshold τ l, th path dpth thrshold τ dpth, and Zipf s thrshold τ Zipf. By xamining th ffctivnss of ach approach, w can choos th bst stting. In our xprimntal procdur, w first ran som prliminary xprimnts to tun th paramtrs of th lmnt filtr and trm filtr. Nxt, with ths tund paramtrs, w masurd th avrag updat tim pr documnt, th indx siz, and th sarch accuracy for ach variation of th proposd mthods. W usd th documnt st with static statistics and chos 50% as th ratio of initial documnts. Finally, w confirmd th ffctivnss of th proposd mthods by using th documnt st with dynamic statistics. B. Prliminary Exprimnts for th Elmnt Filtr and Trm Filtr Th lmnt filtr liminats th lmnts that hav xtrmly short lmnts, xtrmly dp path xprssions, and rarly apparing path xprssions. In this sction, w dscrib som xprimnts that w conductd to dcid th thrsholds. According to th rsults listd in Tabl III, w st τ l to 35; namly, in trms of sarch accuracy, th bst valu for th lmnt-lngth thrshold is 35. Tabl IV shows th proportion of lmnts whos dpth of path xprssions is lss than or qual to τ dpth. W masurd th proportion for all lmnts in th tst collction and for only thos highly rankd lmnts obtaind in our prvious TABLE VI. EFFECTS OF THE PROPOSED APPROACHES updat tim disk run ID (ms/doc) siz(gb) ip[.01] MAiP no-updat (42.1) baslin ST BT OT τ l τ dpth τ Zipf lm filtr trm filtr two filtrs study [8]. Thr is a diffrnc btwn th rsult for all lmnts and that for highly rankd lmnts. This indicats that w can xtract purly usful lmnts if th dpth thrshold is st to xtract as many highly rankd lmnts as possibl and to discard uslss lmnts. W st τ dpth to 6, which ignors any lmnt whos dpth is six or mor. W also invstigatd th thrshold of Zipf s law. W computd mdian frquncy of th path xprssions by using Eq. 7. W st τ Zipf to 166, which ignors any lmnt whos path xprssion appars 166 or fwr tims in th initial indx. In analogy with th lmnt filtr, th trm filtr liminats trms whos wights ar blow th thrshold. W conductd an xprimnt to dcid th thrshold for th trm filtr, as shown in Tabl V. W st n to or τ tw, which ignors th trms whos wights ar lss than th 10,000th largst wight of ach pair of tag and trm. C. Evaluations of th Documnt St with Static Statistics W masurd th avrag updat tim pr documnt, th siz of indics, and th sarch accuracy with ach variation of th proposd mthods, as indicatd in Tabl VI. Not that in th cas of no-updat, or constructing a nw indx from scratch, th avrag updat tim rplacs th construction tim of th initial indics. Not that lmnts whos lngth is lss than τ l ar rmovd from all rsults, vn thos of no-updat. Compard with th ip[.01] of th baslin systm, thos of ST, BT, and OT ar improvd. In particular, ST is th most ffctiv mthod for calculating accurat global wights and is 2.57% mor accurat than th baslin. In addition, th updat fficincis of ths mttjhods ar almost qual. All componnts of th lmnt filtr (i.. τ l, τ dpth, and τ Zipf ) sav updat cost without rducing sarch accuracy. Th combination of τ l, τ dpth, and τ Zipf is th most ffctiv of all possibl combinations and yilds 23.6% fastr updats than th baslin approach. W usd this stting for th lmnt filtr in th subsqunt xprimnts. Th trm filtr also rducs th updat cost by 6.70% without sacrificing sarch accuracy compard with th baslin approach. Nxt, w valuatd th combination of th two filtrs. This approach prforms bttr than ithr of singl filtrs in trms

10 TABLE VII. CATEGORY AND QUERY Catgory nam CQ CW Tchnology and applid scincs Cultur and th arts Natural and physical scincs 9 24 Socity and social scincs 4 13 History and vnts 4 11 Philosophy and thinking 3 8 Gnral rfrnc 3 7 Halth and fitnss 2 7 Popl and slf 3 6 Gography and placs 2 5 Mathmatics and logic 0 0 Rligion and blif systms 0 0 mchan, mtadata, min, motor, musum, ntwork, nikola, opn, opr, park, patnt, program, raid, rcord, rtriv, rotari, scur, social, sourc, storag, systm, tata, tsla, virtual, wirlss CW of cultur : acquisit, africa, al, basktbal, brbr, bilingu, childbirth, childrn, classic, countri, cultur, danc, dish, urop, uropan, fiction, film, food, franc, gam, guitar, hors, instrumnt, japans, kyboard, languag, mahlr, musum, nba, north, prson, picasso, playr, portugus, produc, rgion, rul, scinc, scrabbl, song, spanish, styl, symphoni, tap, tast, trracotta, tradit, typic, vgtarian, vodka, win of both updat fficincy and sarch accuracy. Th updat fficincy is improvd by 26.3%. Th formr xprimnts showd that th sarch accuracy improvd with th path xprssion intgrating mthod and th updat fficincy improvd with two filtrs. Thn, w combind ST and th two filtrs as ST filtrs. Th sarch accuracy improvd by 3.73% compard with th baslin, whil th updat fficincy improvd by 24.9%. In trms of qury fficincy, ach mthod taks 1.5 s to 2.0 s pr qury. This should b accptabl for usrs. Finally, w can attain fast incrmntal updats of indics with an ffctiv and fficint sarch. D. Evaluations of th Documnt St with Dynamic Statistics In th prvious valuations, w assumd that th trm distribution and trm statistics ar static. Howvr, nw topics can mrg suddnly on th Wb and may chang th trm distribution drastically. Hr w artificially assmbl a documnt st with dynamic statistics to invstigat th ffctivnss of th proposd mthods. In this st, th initial documnts do not includ a crtain topic but th updatd documnts do includ th topic. W outlin th stps to valuat as follows: (1) idntify documnts on a crtain topic, (2) construct th initial indx using th othr documnts, and (3) updat th indics incrmntally using th documnts rlatd to th topic. W utilizd th catgoris in Wikipdia to judg whthr a documnt blongs to a crtain topic. Wikipdia has many catgoris of various sizs: twlv major catgoris ar listd in Tabl VII. W sparatd 68 quris into th twlv catgoris. Each qury contains from on to fiv qury kywords, and w obtaind a kyword st for ach catgory. Sinc th catgoris Tchnology and applid scincs (tchnology for short) and Cultur and th arts (cultur for short) includ rlativly larg numbrs of quris (catgory quris, or CQs) and qury kywords (catgory kywords, or CW), w usd ths catgoris in th valuation. W assignd a documnt to a crtain catgory if th documnt contains th catgory kywords. Not that ths catgory kywords ar stmmd. CW of tchnology : aircraft, applid, automobil, aviat, bay, bltchli, brak, car, cod, colossu, compani, comput, databas, dtct, ngin, xprt, fil, filtr, format, graphic, imag, inform, instal, intrus, invnt, java, languag, linux, manag, W usd th catgory quris only to xamin th ffctivnss of th proposd mthods, bcaus w focus on th ffcts of trm distributions with dynamically changing statistics. In this situation, w assumd that usrs xpct an ffctiv sarch to b availabl as soon as nw topics ar addd to th collction. Th numbrs of documnts in th initial indics of tchnology and cultur ar 280,000 and 200,000, rspctivly. W valuatd th ffcts of th changing statistics at four points during th updats. Aftr th updats, th numbr of indxd documnts rachd 660,000 for both catgoris. Tabl VIII lists th ip[.01] of ach catgory for th baslin and ST filtrs. For both catgoris, th proposd mthods attaind bttr sarch accuracis than th baslin. In particular, ST filtrs incrasd th sarch accuracis rapidly vn whn th numbr of updat documnts was small. VII. CONCLUSION In this papr, w proposd mthods for fast incrmntal updats of indics for XML lmnt rtrival to attain both ffctivnss and fficincy in th qury procssing. Th simpl solution for incrmntal updats has two problms: (1) dcrasd sarch accuracy, and (2) incrasd updat tim. W solvd ths problms by intgrating path xprssions and utilizing two filtrs for xcluding unncssary data. Th xprimntal valuations showd that our proposd approachs ar ffctiv and fficint for both static statistics and dynamic statistics. In particular, a variation of th proposd approachs can rduc updat tim by 25% whil th sarch accuracy improvd by 4% compard with th simpl incrmntal updat systm for static statistics. REFERENCES [1] Fi Chn, Xixuan Fng, Christophr Ré, and Min Wang. Optimizing Statistical Information Extraction Programs Ovr Evolving Txt. In Proc. of th 28th IEEE ICDE, [2] Shlomo Gva, Jaap Kamps, and Andrw Trotman. Advancs in Focusd Rtrival. Springr Brlin, [3] Knji Hatano, Hiroko Kinutani, Toshiyuki Amagasa, Yasuhiro Mori, Masatoshi Yoshikawa, and Shunsuk Umura. Analyzing th Proprtis of XML Fragmnts Dcomposd from th INEX Documnt Collction. In Advancs in XML Information Rtrival, volum 3493 of LNCS, pags Springr Brlin, [4] Fang Huang, Stuart Watt, David Harpr, and Malcolm Clark. Compact Rprsntations in XML Rtrival. In Formal Proc. of INEX 2006 Workshop, volum 5631 of LNCS, 2007.

11 TABLE VIII. EFFECTS ON EMERGING A NEW TOPIC # of indxd tchnology (ip[.01]) # of indxd cultur (ip[.01]) doc. ( 10 4 doc.) baslin ST filtrs doc. ( 10 4 doc.) baslin ST filtrs 37 (25% updatd) (25% updatd) (50% updatd) (50% updatd) (75% updatd) (75% updatd) (100% updatd) (100% updatd) [5] Yu Huang, Ziyang Liu, and Yi Chn. Qury Biasd Snippt Gnration in XML Sarch. In Proc. of ACM SIGMOD, pags ACM, [6] Ihab F. Ilyas, Gorg Bskals, and Mohamd A. Soliman. A Survy of Top-k Qury Procssing Tchniqus in Rlational Databas Systms. ACM Computing Survys (CSUR), 40:1 58, [7] Atsushi Kyaki, Knji Hatano, and Jun Miyazaki. Rlaxd Global Trm Wights for XML Elmnt Sarch. In Formal Proc. of INEX 2010 Workshop, volum 6932 of LNCS, [8] Atsushi Kyaki, Knji Hatano, and Jun Miyazaki. Rsult Rconstruction Approach for Mor Effctiv XML Elmnt Sarch. Intrnational Journal of Wb Information Systms (IJWIS), 7(4): , [9] Atsushi Kyaki, Jun Miyazaki, Knji Hatano, Goshiro Yamamoto, Takafumi Taktomi, and Hirokazu Kato. Path Exprssion-basd Smoothing of Qury Liklihood Modl for XML Elmnt Rtrival. In Proc. of th 1st ACIS Intrnational Symposium on Applid Computing & Information Tchnology, 2013 (to appar). [10] Nicholas Lstr, Justin Zobl, and Hugh E. Williams. In-Plac vrsus R-Build vrsus R-Mrg: Indx Maintnanc Stratgis for Txt Rtrival Systms. In Proc. of th 27th Australasian confrnc on Computr Scinc, [11] Fang Liu, Clmnt Yu, Wiyi Mng, and Abdur Chowdhury. Effctiv Kyword sarch in Rlational Databass. In Proc. of ACM SIGMOD, [12] Wi Liu, Stphn Robrtson, and Andrw Macfarlan. Fild-Wightd XML Rtrival Basd on BM25. In Formal Proc. of INEX 2005 Workshop, volum 3977 of LNCS, [13] Ziyang Liu and Yi Chn. Idntifying Maningful Rturn Information for XML Kyword Sarch. In Proc. of ACM SIGMOD, pags ACM, [14] Christophr D. Manning, Prabhakar Raghavan, and Hinrich Schutz. Introduction to Information Rtrival, pags Cambridg Univrsity Prss, [15] Giorgos Margaritis and Strgios V. Anastasiadis. Low-cost Managmnt of Invrtd Fils for Onlin Full-Txt Sarch. In Proc. of 18th ACM CIKM, [16] M. E. Maron. Automatic Indxing: An Exprimntal Inquiry. Journal of th ACM, 8: , [17] Thomas Numann and Grhard Wikum. xrdf3x: Fast Qurying, High Updat Rats, and Consistncy for RDF Databass. In Proc. of 36th VLDB, pags , [18] Bnjamin Piwowarski and Patrick Gallinari. A Baysian Framwork for XML Information Rtrival: Sarching and Larning with th INEX Collction. Journal of Information Rtrival, 8(4): , [19] Chnghui Rn, Eric Lo, Bn Kao, Xinji Zhu, and Rynold Chng. On Qurying Historical Evolving Graph Squncs. In Proc. of th 37th VLDB, [20] Martin Thobald, Holgr Bast, Dbapriyo Majumdar, Ralf Schnkl, and Grhard Wikum. TopX: Efficint and Vrsatil Top-k Qury Procssing for Smistructurd Data. Th VLDB Journal, 17(1):81 115, [21] Anthony Tomasic, Héctor García-Molina, and Kurt Shons. Incrmntal Updats of Invrtd Lists for Txt Documnt Rtrival. In Proc. of ACM SIGMOD, [22] Andrw Trotman, Xiang-Fi Jia, and Shlomo Gva. Fast and Effctiv Focusd Rtrival. In Formal Proc. of INEX 2009 Workshop, volum 6203 of LNCS, [23] Andrw Trotman and Börkur Sigurbjörnsson. Narrowd Extndd XPath I (NEXI). In Formal Proc. of INEX 2004 Workshop, volum 3493 of LNCS, 2005.

Objectives. Two Ways to Implement Lists. Lists. Chapter 24 Implementing Lists, Stacks, Queues, and Priority Queues

Objectives. Two Ways to Implement Lists. Lists. Chapter 24 Implementing Lists, Stacks, Queues, and Priority Queues Chaptr 24 Implmnting Lists, Stacks, Quus, and Priority Quus CS2: Data Structurs and Algorithms Colorado Stat Univrsity Original slids by Danil Liang Modifid slids by Chris Wilcox Objctivs q To dsign common

More information

CSE 272 Assignment 1

CSE 272 Assignment 1 CSE 7 Assignmnt 1 Kui-Chun Hsu Task 1: Comput th irradianc at A analytically (point light) For point light, first th nrgy rachd A was calculatd, thn th nrgy was rducd by a factor according to th angl btwn

More information

Systems in Three Variables. No solution No point lies in all three planes. One solution The planes intersect at one point.

Systems in Three Variables. No solution No point lies in all three planes. One solution The planes intersect at one point. 3-5 Systms in Thr Variabls TEKS FOCUS VOCABULARY TEKS (3)(B) Solv systms of thr linar quations in thr variabls by using Gaussian limination, tchnology with matrics, and substitution. Rprsntation a way

More information

2018 How to Apply. Application Guide. BrandAdvantage

2018 How to Apply. Application Guide. BrandAdvantage 2018 How to Apply Application Guid BrandAdvantag Contnts Accssing th Grant Sit... 3 Wlcom pag... 3 Logging in To Pub Charity... 4 Rgistration for Nw Applicants ( rgistr now )... 5 Organisation Rgistration...

More information

Midterm 2 - Solutions 1

Midterm 2 - Solutions 1 COS 26 Gnral Computr Scinc Spring 999 Midtrm 2 - Solutions. Writ a C function int count(char s[ ]) that taks as input a \ trminatd string and outputs th numbr of charactrs in th string (not including th

More information

The Size of the 3D Visibility Skeleton: Analysis and Application

The Size of the 3D Visibility Skeleton: Analysis and Application Th Siz of th 3D Visibility Sklton: Analysis and Application Ph.D. thsis proposal Linqiao Zhang lzhang15@cs.mcgill.ca School of Computr Scinc, McGill Univrsity March 20, 2008 thsis proposal: Th Siz of th

More information

Problem Set 1 (Due: Friday, Sept. 29, 2017)

Problem Set 1 (Due: Friday, Sept. 29, 2017) Elctrical and Computr Enginring Mmorial Univrsity of Nwfoundland ENGI 9876 - Advancd Data Ntworks Fall 2017 Problm St 1 (Du: Friday, Spt. 29, 2017) Qustion 1 Considr a communications path through a packt

More information

The semantic WEB Roles of XML & RDF

The semantic WEB Roles of XML & RDF Th smantic WEB Rols of XML & RDF STEFAN DECKER AND SERGEY MELNIK FRANK VAN HARMELEN, DIETER FENSEL, AND MICHEL KLEIN JEEN BROEKSTRA MICHAEL ERDMANN IAN HORROCKS Prsntd by: Iniyai Thiruvalluvan CSCI586

More information

Managing Trust Relationships in Peer 2 Peer Systems

Managing Trust Relationships in Peer 2 Peer Systems Managing Trust Rlationships in Pr 2 Pr Systms R.S.SINJU PG STUDENT, DEPARTMENT OF COMPUTER SCIENCE, PONJESLY COLLEGE OF ENGINEERING NAGERCOIL, TAMILNADU, INDIA C.FELSY ASST.PROF, DEPARTMENT OF COMPUTER

More information

Register Allocation. Register Allocation

Register Allocation. Register Allocation Rgistr Allocation Jingk Li Portlan Stat Univrsity Jingk Li (Portlan Stat Univrsity) CS322 Rgistr Allocation 1 / 28 Rgistr Allocation Assign an unboun numbr of tmporaris to a fix numbr of rgistrs. Exampl:

More information

Figure 1: XML document Figure 3: XML element Figure 2: XML tree class to handle this. Such a method would be especially useful as new topics are added

Figure 1: XML document Figure 3: XML element Figure 2: XML tree class to handle this. Such a method would be especially useful as new topics are added Fast and Incremental Indexing in Effective and Efficient XML Element Retrieval Systems Atsushi Keyaki, Jun Miyazaki Graduate School of Information Science Nara Institute of Science and Technology 8916-5

More information

Reimbursement Requests in WORKS

Reimbursement Requests in WORKS Rimbursmnt Rqusts in WORKS Important points about Rimbursmnts in Works Rimbursmnt Rqust is th procss by which UD mploys will b rimbursd for businss xpnss paid using prsonal funds. Rimbursmnt Rqust can

More information

Intersection-free Dual Contouring on Uniform Grids: An Approach Based on Convex/Concave Analysis

Intersection-free Dual Contouring on Uniform Grids: An Approach Based on Convex/Concave Analysis Intrsction-fr Dual Contouring on Uniform Grids: An Approach Basd on Convx/Concav Analysis Charli C. L. Wang Dpartmnt of Mchanical and Automation Enginring, Th Chins Univrsity of Hong Kong E-mail: cwang@ma.cuhk.du.hk

More information

Lesson Focus: Finding Equivalent Fractions

Lesson Focus: Finding Equivalent Fractions Lsson Plans: Wk of 1-26-15 M o n Bindrs: /Math;; complt on own, thn chck togthr Basic Fact Practic Topic #10 Lsson #5 Lsson Focus: Finding Equivalnt Fractions *Intractiv Larning/Guidd Practic-togthr in

More information

A New Algorithm for Solving Shortest Path Problem on a Network with Imprecise Edge Weight

A New Algorithm for Solving Shortest Path Problem on a Network with Imprecise Edge Weight Availabl at http://pvamudu/aam Appl Appl Math ISSN: 193-9466 Vol 6, Issu (Dcmbr 011), pp 60 619 Applications and Applid Mathmatics: An Intrnational Journal (AAM) A Nw Algorithm for Solving Shortst Path

More information

Formal Foundation, Approach, and Smart Tool for Software Models Comparison

Formal Foundation, Approach, and Smart Tool for Software Models Comparison Formal Foundation, Approach, and Smart Tool for Softwar Modls Comparison Olna V. Chbanyuk, Abdl-Badh M. Salm Softwar Enginring Dpartmnt, National Aviation Univrsity, Kyiv, Ukrain Computr Scinc, Faculty

More information

8.3 INTEGRATION BY PARTS

8.3 INTEGRATION BY PARTS 8.3 Intgration By Parts Contmporary Calculus 8.3 INTEGRATION BY PARTS Intgration by parts is an intgration mthod which nabls us to find antidrivativs of som nw functions such as ln(x) and arctan(x) as

More information

Parser Self-Training for Syntax-Based Machine Translation

Parser Self-Training for Syntax-Based Machine Translation arsr Slf-Training for Syntax-Basd Machin Translation Makoto Morishita, Koichi Akab, Yuto Hatakoshi Graham ubig, Koichiro Yoshino, Satoshi akamrua Graduat School of Information Scinc ara Institut of Scinc

More information

1. Trace the array for Bubble sort 34, 8, 64, 51, 32, 21. And fill in the following table

1. Trace the array for Bubble sort 34, 8, 64, 51, 32, 21. And fill in the following table 1. Trac th array for Bubbl sort 34, 8, 64, 51, 3, 1. And fill in th following tabl bubbl(intgr Array x, Intgr n) Stp 1: Intgr hold, j, pass; Stp : Boolan switchd = TRUE; Stp 3: for pass = 0 to (n - 1 &&

More information

Fequent Pattern Recognization From Stream Data Using Compact Data Structure

Fequent Pattern Recognization From Stream Data Using Compact Data Structure Fqunt Pattrn Rcognization From Stram Data Using Compact Data Structur Fabin M Christian 1, Narndra C.Chauhan 2, Nilsh B. Prajapati 3 1 PG Scholar, CE Dpartmnt, BVM Engg. Collg, V.V.Nagar, fabin.christian@gmail.com

More information

i e ai E ig e v / gh E la ES h E A X h ES va / A SX il E A X a S

i e ai E ig e v / gh E la ES h E A X h ES va / A SX il E A X a S isto C o C or Co r op ra p a py ag yr g ri g g gh ht S S S V V K r V K r M K v M r v M rn v MW n W S r W Sa r W K af r: W K f : a H a M r T H r M rn w T H r Mo ns w T i o S ww c ig on a w c g nd af ww

More information

AN EVALUATION MODEL FOR THE CHAINS OF DISTRIBUTED MULTIMEDIA INDEXING TOOLS RESPECTING USER PREFERENCES

AN EVALUATION MODEL FOR THE CHAINS OF DISTRIBUTED MULTIMEDIA INDEXING TOOLS RESPECTING USER PREFERENCES AN EVALUATION MODEL FOR THE CHAINS OF DISTRIBUTED MULTIMEDIA INDEXING TOOLS RESPECTING USER PREFERENCES 1 Bassm HAIDAR, 2 Bilal CHEBARO, 3 Hassan WEHBI 1 Asstt Prof., Dpartmnt of Computr Scincs, Faculty

More information

TCP Congestion Control. Congestion Avoidance

TCP Congestion Control. Congestion Avoidance TCP Congstion Control TCP sourcs chang th snding rat by modifying th window siz: Window = min {Advrtisd window, Congstion Window} Rcivr Transmittr ( cwnd ) In othr words, snd at th rat of th slowst componnt:

More information

Comment (justification for change) by the MB

Comment (justification for change) by the MB Editor's disposition s CD2 19763-12 as at 2013-11-03 Srial Annx (.g. 3.1) Figur/ Tabl/t (.g. Tabl 1) 001 CA 00 All All - G Canada disapprovs th draft for th rasons blow. 002 GB 01 Gnral d numbring has

More information

To Do. Mesh Data Structures. Mesh Data Structures. Motivation. Outline. Advanced Computer Graphics (Fall 2010) Desirable Characteristics 1

To Do. Mesh Data Structures. Mesh Data Structures. Motivation. Outline. Advanced Computer Graphics (Fall 2010) Desirable Characteristics 1 Advancd Computr Graphics (Fall 200) CS 283, Lctur 5: Msh Data Structurs Ravi Ramamoorthi http://inst.cs.brkly.du/~cs283/fa0 To Do Assignmnt, Du Oct 7. Start rading and working on it now. Som parts you

More information

To Do. Advanced Computer Graphics. Motivation. Mesh Data Structures. Outline. Mesh Data Structures. Desirable Characteristics 1

To Do. Advanced Computer Graphics. Motivation. Mesh Data Structures. Outline. Mesh Data Structures. Desirable Characteristics 1 Advancd Computr Graphics CSE 63 [Spring 207], Lctur 7 Ravi Ramamoorthi http://www.cs.ucsd.du/~ravir To Do Assignmnt, Du Apr 28 Any last minut issus or difficultis? Starting Gomtry Procssing Assignmnt 2

More information

A Brief Summary of Draw Tools in MS Word with Examples! ( Page 1 )

A Brief Summary of Draw Tools in MS Word with Examples! ( Page 1 ) A Brif Summary of Draw Tools in MS Word with Exampls! ( Pag 1 ) Click Viw command at top of pag thn Click Toolbars thn Click Drawing! A chckmark appars in front of Drawing! A toolbar appars at bottom of

More information

Recorder Variables. Defining Variables

Recorder Variables. Defining Variables Rcordr Variabls Dfining Variabls Simpl Typs Complx Typs List of Rsrvd Words Using Variabls Stting Action Paramtrs Parsing Lists and Tabls Gtting Valu from Lists and Tabls Using Indxs with Lists Using Indxs

More information

XML Publisher with connected query: A Primer. Session #30459 March 19, 2012

XML Publisher with connected query: A Primer. Session #30459 March 19, 2012 XML Publishr with connctd qury: A Primr Sssion #30459 March 19, 2012 Agnda/ Contnts Introduction Ovrviw of XMLP Gtting Startd Bst practics for building a basic XMLP rport Connctd Qury Basics Building a

More information

Terrain Mapping and Analysis

Terrain Mapping and Analysis Trrain Mapping and Analysis Data for Trrain Mapping and Analysis Digital Trrain Modl (DEM) DEM rprsnts an array of lvation points. Th quality of DEM influncs th accuracy of trrain masurs such as slop and

More information

TRIANGULATION OF NURBS SURFACES. Jamshid Samareh-Abolhassani. 1 Abstract

TRIANGULATION OF NURBS SURFACES. Jamshid Samareh-Abolhassani. 1 Abstract TRIANGULATION OF NURBS SURFACES Jamshid Samarh-Abolhassani 1 Abstract A tchniqu is prsntd for triangulation of NURBS surfacs. This tchniqu is built upon an advancing front tchniqu combind with grid point

More information

An Agent-Based Architecture for Service Discovery and Negotiation in Wireless Networks

An Agent-Based Architecture for Service Discovery and Negotiation in Wireless Networks An Agnt-Basd Architctur for Srvic Discovry and Ngotiation in Wirlss Ntworks Abstract Erich Birchr and Torstn Braun Univrsity of Brn, Nubrückstrass 10, 3012 Brn, Switzrland Email: braun@iam.unib.ch This

More information

SPECKLE NOISE REDUCTION IN SAR IMAGING USING 2-D LATTICE FILTERS BASED SUBBAND DECOMPOSITION

SPECKLE NOISE REDUCTION IN SAR IMAGING USING 2-D LATTICE FILTERS BASED SUBBAND DECOMPOSITION 7th Europan Signal Procssing Confrnc EUSIPCO 9 Glasgow Scotland August 4-8 9 SPECKLE REDUCTION IN SAR IMAGING USING -D LATTICE FILTERS ASED SUAND DECOMPOSITION Göhan Karasaal N.. Kaplan I. Err Informatics

More information

: Mesh Processing. Chapter 6

: Mesh Processing. Chapter 6 600.657: Msh Procssing Chaptr 6 Quad-Dominant Rmshing Goal: Gnrat a rmshing of th surfac that consists mostly of quads whos dgs align with th principal curvatur dirctions. [Marinov t al. 04] [Alliz t al.

More information

" dx v(x) $ % You may also have seen this written in shorthand form as. & ' v(x) + u(x) '# % ! d

 dx v(x) $ % You may also have seen this written in shorthand form as. & ' v(x) + u(x) '# % ! d Calculus II MAT 146 Mthods of Intgration: Intgration by Parts Just as th mthod of substitution is an intgration tchniqu that rvrss th drivativ procss calld th chain rul, Intgration by parts is a mthod

More information

Evolutionary Clustering and Analysis of Bibliographic Networks

Evolutionary Clustering and Analysis of Bibliographic Networks Evolutionary Clustring and Analysis of Bibliographic Ntworks Manish Gupta Univrsity of Illinois at Urbana-Champaign gupta58@illinois.du Charu C. Aggarwal IBM T. J. Watson Rsarch Cntr charu@us.ibm.com Jiawi

More information

Installation Saving. Enhanced Physical Durability Enhanced Performance Warranty The IRR Comparison

Installation Saving. Enhanced Physical Durability Enhanced Performance Warranty The IRR Comparison Contnts Tchnology Nwly Dvlopd Cllo Tchnology Cllo Tchnology : Improvd Absorption of Light Doubl-sidd Cll Structur Cllo Tchnology : Lss Powr Gnration Loss Extrmly Low LID Clls 3 3 4 4 4 Advantag Installation

More information

Shift. Reduce. Review: Shift-Reduce Parsing. Bottom-up parsing uses two actions: Bottom-Up Parsing II. ABC xyz ABCx yz. Lecture 8.

Shift. Reduce. Review: Shift-Reduce Parsing. Bottom-up parsing uses two actions: Bottom-Up Parsing II. ABC xyz ABCx yz. Lecture 8. Rviw: Shift-Rduc Parsing Bottom-up parsing uss two actions: Bottom-Up Parsing II Lctur 8 Shift ABC xyz ABCx yz Rduc Cbxy ijk CbA ijk Prof. Aikn CS 13 Lctur 8 1 Prof. Aikn CS 13 Lctur 8 2 Rcall: h Stack

More information

Linked Data meet Sensor Networks

Linked Data meet Sensor Networks Digital Entrpris Rsarch Institut www.dri.i Linkd Data mt Snsor Ntworks Myriam Lggiri DERI NUI Galway, Irland Copyright 2008 Digital Entrpris Rsarch Institut. All rights rsrvd. Linkd Data mt Snsor Ntworks

More information

Clustering Belief Functions using Extended Agglomerative Algorithm

Clustering Belief Functions using Extended Agglomerative Algorithm IJ Imag Graphics and Signal Procssing 0 - Publishd Onlin Fbruary 0 in MECS (http://wwwmcs-prssorg/ ing Blif Functions using Extndd Agglomrativ Algorithm Ying Png Postgraduat Collg Acadmy of Equipmnt Command

More information

Efficient Obstacle-Avoiding Rectilinear Steiner Tree Construction

Efficient Obstacle-Avoiding Rectilinear Steiner Tree Construction Efficint Obstacl-Avoiding Rctilinar Stinr Tr Construction Chung-Wi Lin, Szu-Yu Chn, Chi-Fng Li, Yao-Wn Chang, and Chia-Lin Yang Graduat Institut of Elctronics Enginring Dpartmnt of Elctrical Enginring

More information

Vignette to package samplingdatacrt

Vignette to package samplingdatacrt Vigntt to packag samplingdatacrt Diana Trutschl Contnts 1 Introduction 1 11 Objctiv 1 1 Diffrnt study typs 1 Multivariat normal distributd data for multilvl data 1 Fixd ffcts part Random part 9 3 Manual

More information

Principles of Programming Languages Topic: Formal Languages II

Principles of Programming Languages Topic: Formal Languages II Principls of Programming Languags Topic: Formal Languags II CS 34,LS, LTM, BR: Formal Languags II Rviw A grammar can b ambiguous i.. mor than on pars tr for sam string of trminals in a PL w want to bas

More information

Interfacing the DP8420A 21A 22A to the AN-538

Interfacing the DP8420A 21A 22A to the AN-538 Intrfacing th DP8420A 21A 22A to th 68000 008 010 INTRODUCTION This application not xplains intrfacing th DP8420A 21A 22A DRAM controllr to th 68000 Thr diffrnt dsigns ar shown and xplaind It is assumd

More information

HEAD DETECTION AND TRACKING SYSTEM

HEAD DETECTION AND TRACKING SYSTEM HEAD DETECTION AND TRACKING SYSTEM Akshay Prabhu 1, Nagacharan G Tamhankar 2,Ashutosh Tiwari 3, Rajsh N(Assistant Profssor) 4 1,2,3,4 Dpartmnt of Information Scinc and Enginring,Th National Institut of

More information

About Notes And Symbols

About Notes And Symbols About Nots And Symbols by Batric Wildr Contnts Sht 1 Sht 2 Sht 3 Sht 4 Sht 5 Sht 6 Sht 7 Sht 8 Sht 9 Sht 10 Sht 11 Sht 12 Sht 13 Sht 14 Sht 15 Sht 16 Sht 17 Sht 18 Sht 19 Sht 20 Sht 21 Sht 22 Sht 23 Sht

More information

Workbook for Designing Distributed Control Applications using Rockwell Automation s HOLOBLOC Prototyping Software John Fischer and Thomas O.

Workbook for Designing Distributed Control Applications using Rockwell Automation s HOLOBLOC Prototyping Software John Fischer and Thomas O. Workbook for Dsigning Distributd Control Applications using Rockwll Automation s HOLOBLOC Prototyping Softwar John Fischr and Thomas O. Bouchr Working Papr No. 05-017 Introduction A nw paradigm for crating

More information

The Network Layer: Routing Algorithms. The Network Layer: Routing & Addressing Outline

The Network Layer: Routing Algorithms. The Network Layer: Routing & Addressing Outline PS 6 Ntwork Programming Th Ntwork Layr: Routing lgorithms Michl Wigl partmnt of omputr Scinc lmson Univrsity mwigl@cs.clmson.du http://www.cs.clmson.du/~mwigl/courss/cpsc6 Th Ntwork Layr: Routing & ddrssing

More information

Dynamic Light Trail Routing and Protection Issues in WDM Optical Networks

Dynamic Light Trail Routing and Protection Issues in WDM Optical Networks This full txt papr was pr rviwd at th dirction of IEEE Communications Socity subjct mattr xprts for publication in th IEEE GLOBECOM 2005 procdings. Dynamic Light Trail Routing and Protction Issus in WDM

More information

Oracle Data Relationship Management Suite User's Guide. Release

Oracle Data Relationship Management Suite User's Guide. Release Oracl Data Rlationship Managmnt Suit Usr's Guid Rlas 11.1.2.4.346 E75912-02 Jun 2018 Oracl Data Rlationship Managmnt Suit Usr's Guid, Rlas 11.1.2.4.346 E75912-02 Copyright 1999, 2018, Oracl and/or its

More information

FSP Synthesis of an off-set five bar-slider mechanism with variable topology

FSP Synthesis of an off-set five bar-slider mechanism with variable topology FSP Synthsis of an off-st fiv bar-slidr mchanism with variabl topology Umsh. M. Daivagna 1*, Shrinivas. S. Balli 2 1 Dpartmnt of Mchanical Enginring, S.T.J.Institut of Tchnology, Ranbnnur, India 2 Dpt.

More information

A Vision-based Navigation System of Mobile Tracking Robot

A Vision-based Navigation System of Mobile Tracking Robot A Vision-basd Navigation Systm of Mobil Tracking Robot Ji Wu Vac1av Snasl Dpt. Computr Scinc FCS VSB - Tchnical Univrsity of Ostrava Ostrava Czch Rpublic dfrmat2008 @hotmail.com vaclav. snasl @vsb.cz Ajith

More information

To Do. Advanced Computer Graphics. Motivation. Mesh Data Structures. Outline. Mesh Data Structures. Desirable Characteristics 1

To Do. Advanced Computer Graphics. Motivation. Mesh Data Structures. Outline. Mesh Data Structures. Desirable Characteristics 1 Advancd Computr Graphics CSE 63 [Spring 208], Lctur 7 Ravi Ramamoorthi http://www.cs.ucsd.du/~ravir To Do Assignmnt, Du Apr 27 Any last minut issus or difficultis? Starting Gomtry Procssing Assignmnt 2

More information

Fuzzy Intersection and Difference Model for Topological Relations

Fuzzy Intersection and Difference Model for Topological Relations IFS-EUSFLT 009 Fuzzy Intrsction and Diffrnc Modl for Topological Rlations hd LOODY Flornc SEDES Jordi INGLD 3 Univrsité Paul Sabatir (UPS) Toulous, 8 Rout d Narbonn, F-306-CEDEX 9, Franc Institut d Rchrchn

More information

CPSC 826 Internetworking. The Network Layer: Routing & Addressing Outline. The Network Layer: Routing Algorithms. Routing Algorithms Taxonomy

CPSC 826 Internetworking. The Network Layer: Routing & Addressing Outline. The Network Layer: Routing Algorithms. Routing Algorithms Taxonomy PS Intrntworking Th Ntwork Layr: Routing & ddrssing Outlin Th Ntwork Layr: Routing lgorithms Michl Wigl partmnt of omputr Scinc lmson Univrsity mwigl@cs.clmson.du Novmbr, Ntwork layr functions Routr architctur

More information

Acceleration of the Smith-Waterman Algorithm using Single and Multiple Graphics Processors

Acceleration of the Smith-Waterman Algorithm using Single and Multiple Graphics Processors Acclration of th Smith-Watrman Algorithm using Singl and Multipl Graphics Procssors Ali Khah-Sad, Stphn Pool and J. Blair Prot Abstract Finding rgions of similarity btwn two vry long data strams is a computationally

More information

Intersection-free Contouring on An Octree Grid

Intersection-free Contouring on An Octree Grid Intrsction-fr Contouring on An Octr Grid Tao Ju Washington Univrsity in St. Louis On Brookings Driv St. Louis, MO 0, USA taoju@cs.wustl.du Tushar Udshi Zyvx Corporation North Plano Road Richardson, Txas

More information

2 Mega Pixel. HD-SDI Bullet Camera. User Manual

2 Mega Pixel. HD-SDI Bullet Camera. User Manual 2 Mga Pixl HD-SDI Bullt Camra Usr Manual Thank you for purchasing our product. This manual is only applicabl to SDI bullt camras. Thr may b svral tchnically incorrct placs or printing rrors in this manual.

More information

An Auto-tuned Method for Solving Large Tridiagonal Systems on the GPU

An Auto-tuned Method for Solving Large Tridiagonal Systems on the GPU An Auto-tund Mthod for Solving Larg Tridiagonal Systms on th GPU Andrw Davidson Univrsity of California, Davis aaldavidson@ucdavis.du Yao Zhang Univrsity of California, Davis yaozhang@ucdavis.du John D.

More information

Spectral sensitivity and color formats

Spectral sensitivity and color formats FirWir camras Spctral snsitivity and color formats At th "input" of a camra, w hav a CCD chip. It transforms photons into lctrons. Th spctral snsitivity of this transformation is an important charactristic

More information

Type & Media Page 1. January 2014 Libby Clarke

Type & Media Page 1. January 2014 Libby Clarke Nam: 1 In ordr to hlp you s your progrss at th nd of this ntir xrcis, you nd to provid som vidnc of your starting point. To start, draw th a on th lft into th box to th right, dpicting th sam siz and placmnt.

More information

Greedy Algorithms. Interval Scheduling. Greedy Algorithm. Optimality. Greedy Algorithm (cntd) Greed is good. Greed is right. Greed works.

Greedy Algorithms. Interval Scheduling. Greedy Algorithm. Optimality. Greedy Algorithm (cntd) Greed is good. Greed is right. Greed works. Algorithm Grdy Algorithm 5- Grdy Algorithm Grd i good. Grd i right. Grd work. Wall Strt Data Structur and Algorithm Andri Bulatov Algorithm Grdy Algorithm 5- Algorithm Grdy Algorithm 5- Intrval Schduling

More information

Probabilistic inference

Probabilistic inference robabilistic infrnc Suppos th agnt has to mak a dcision about th valu of an unobsrvd qury variabl X givn som obsrvd vidnc E = artially obsrvabl, stochastic, pisodic nvironmnt Eampls: X = {spam, not spam},

More information

Dynamic Spatial Partitioning for Real-Time Visibility Determination

Dynamic Spatial Partitioning for Real-Time Visibility Determination Dynamic Spatial Partitioning for Ral-Tim Visibility Dtrmination Joshua Shagam Josph J. Pfiffr, Jr. Nw Mxico Stat Univrsity Abstract Th static spatial partitioning mchanisms usd in currnt intractiv systms,

More information

EXTENSION OF RCC TOPOLOGICAL RELATIONS FOR 3D COMPLEX OBJECTS COMPONENTS EXTRACTED FROM 3D LIDAR POINT CLOUDS

EXTENSION OF RCC TOPOLOGICAL RELATIONS FOR 3D COMPLEX OBJECTS COMPONENTS EXTRACTED FROM 3D LIDAR POINT CLOUDS Th Intrnational rchivs of th Photogrammtry, mot Snsing and Spatial Information Scincs, Volum XLI-, 016 XXIII ISPS Congrss, 1 19 July 016, Pragu, Czch public EXTENSION OF CC TOPOLOGICL ELTIONS FO D COMPLEX

More information

Tillförlitlig dimensionering mot utmattning UTMIS Vårmöte 2018 på Högskolan i Skövde

Tillförlitlig dimensionering mot utmattning UTMIS Vårmöte 2018 på Högskolan i Skövde Tillförlitlig dimnsionring mot utmattning UTMIS Vårmöt 2018 på Högskolan i Skövd Rami Mansour & Mårtn Olsson KTH Hållfasthtslära mart@kth.s ramimans@kth.s Introduction Ovrviw of rliabl dsign Traditional

More information

running at 133 MHz bus. A Pentium III 1.26GHz with 512K cache running at 133 MHz bus is an available option. Fits Your Needs

running at 133 MHz bus. A Pentium III 1.26GHz with 512K cache running at 133 MHz bus is an available option. Fits Your Needs 3715 Industrial PCs 15.0" LCD Flat Panl Display DS-371500(E) Xycom Automation's nwst gnration of Industrial PCs is dsignd and tstd for th tough nvironmnts rquird for plant floor us. Our standard PC configurations

More information

Extending z/tpf using IBM API Management (APIM)

Extending z/tpf using IBM API Management (APIM) Extnding using API Managmnt (APIM) Mark Gambino, TPF Dvlopmnt Lab March 23, 2015 TPFUG Dallas, TX Th Big Pictur Goal Mobil Applications Cloud APIs Cloud-basd Srvics On-Prmis Entrpris APIs E n t r p r I

More information

Adaptive subband selection in OFDM-based cognitive radios for better system coexistence

Adaptive subband selection in OFDM-based cognitive radios for better system coexistence Univrsity of Wollongong Rsarch Onlin Faculty of Informatics - Paprs (Archiv) Faculty of Enginring and Information Scincs 28 Adaptiv subband slction in OFDM-basd cognitiv radios for bttr systm coxistnc

More information

Summary: Semantic Analysis

Summary: Semantic Analysis Summary: Smantic Analysis Chck rrors not dtctd by lxical or syntax analysis Intrmdiat Cod Scop rrors: Variabls not dfind Multipl dclarations Typ rrors: Assignmnt of valus of diffrnt typs Invocation of

More information

Reliability Coordinator Base Schedule Aggregation Portal (RC BSAP) Interface Specification for RC BSAP Services

Reliability Coordinator Base Schedule Aggregation Portal (RC BSAP) Interface Specification for RC BSAP Services Rliability Coordinator Bas Schdul Aggrgation Portal (RC BSAP) Intrfac Spcification for RC BSAP Srvics (Businss Ruls v 10.x(Spring 2019) or latr) Vrsion: 1.1 vmbr 6, 2018 Rvision History Dat Vrsion By Dscription

More information

Dual-mode Operation of the Finger-type Manipulator Based on Distributed Actuation Mechanism

Dual-mode Operation of the Finger-type Manipulator Based on Distributed Actuation Mechanism 11 th World Congrss on Structural and Multidisciplinary Optimisation 07 th -1 th, Jun 015, Sydny Australia Dual-mod Opration of th Fingr-typ Manipulator Basd on Distributd Actuation Mchanism Jong Ho Kim

More information

Building a Scanner, Part I

Building a Scanner, Part I COMP 506 Ric Univrsity Spring 2018 Building a Scannr, Part I sourc cod IR Front End Optimizr Back End IR targt cod Copyright 2018, Kith D. Coopr & Linda Torczon, all rights rsrvd. Studnts nrolld in Comp

More information

Misbehavior in Nash Bargaining Solution Allocation

Misbehavior in Nash Bargaining Solution Allocation Misbhavior in Nash Bargaining Solution Allocation Ilya Nikolavskiy, Andry Lukyannko, Andri Gurtov Aalto Univrsity, Finland, firstnam.lastnam@aalto.fi Hlsinki Institut for Information Tchnology, Finland,

More information

An Architecture for Hierarchical Collision Detection

An Architecture for Hierarchical Collision Detection An Architctur for Hirarchical Collision Dtction Gabril Zachmann Computr Graphics, Informatik II Univrsity of Bonn mail: zach@cs.uni-bonn.d Güntr Knittl WSI/GRIS Univrsity of Tübingn mail: knittl@gris.uni-tubingn.d

More information

Usage of Ontology-Based Semantic Analysis of Complex Information Objects in Virtual Research Environments

Usage of Ontology-Based Semantic Analysis of Complex Information Objects in Virtual Research Environments Usag of Ontology-Basd Smantic Analysis of Complx Information Objcts in Virtual Rsarch Environmnts Julia Rogushina 1, Anatoly Gladun 2, Abdl-Badh M. Salm 3 1 Institut of Softwar Systms of National Acadmy

More information

Bit-array (4096kbit) Logic. Reconfigurable. Column-select. Row-select

Bit-array (4096kbit) Logic. Reconfigurable. Column-select. Row-select Low-Powr Dsign of Pag-Basd Intllignt Mmory Mark Oskin, Frdric T. Chong, Aamir Farooqui, Timothy Shrwood, and Justin Hnsly Univrsity of California at Davis Abstract Advancs in DRAM tchnology hav ld many

More information

CC-RANSAC: Fitting Planes in the Presence of Multiple Surfaces in Range Data

CC-RANSAC: Fitting Planes in the Presence of Multiple Surfaces in Range Data CC-RANSAC: Fitting Plans in th Prsnc of Multipl Surfacs in Rang Data Orazio Gallo, Robrto Manduchi Univrsity of California, Santa Cruz Abbas Rafii Cansta, Inc. Abstract Rang snsors, in particular tim-of-flight

More information

Review of Different Histogram Equalization Based Contrast Enhancement Techniques

Review of Different Histogram Equalization Based Contrast Enhancement Techniques Intrnational Journal of Advancd Rsarch in Computr and Communication Enginring Vol. 3, Issu 7, July 24 ISSN (Onlin) : 2278-2 Rviw of Diffrnt Histogram Equalization Basd Contrast Enhancmnt Tchniqus Er. Shfali

More information

Introduction to Data Mining

Introduction to Data Mining Introduction to Data Mining Lctur #15: Clustring-2 Soul National Univrsity 1 In Tis Lctur Larn t motivation and advantag of BFR, an xtnsion of K-mans to vry larg data Larn t motivation and advantag of

More information

Lightweight Polymorphic Effects

Lightweight Polymorphic Effects Lightwight Polymorphic Effcts Lukas Rytz, Martin Odrsky, and Philipp Hallr EPFL, Switzrland, {first.last}@pfl.ch Abstract. Typ-and-ffct systms ar a wll-studid approach for rasoning about th computational

More information

Performance Analysis of IEEE MAC Protocol with Different ACK Polices

Performance Analysis of IEEE MAC Protocol with Different ACK Polices Prformanc Analysis of IEEE 82.15.3 MAC Protocol with Diffrnt Polics S. Mhta and K.S. Kwak Wirlss Communications Rsarch Cntr, Inha Univrsity, Kora suryanand.m@gmail.com Abstract. h wirlss prsonal ara ntwork

More information

XML security in certificate management

XML security in certificate management XML scurity in crtificat managmnt Joan Lu, Nathan Cripps and Chn Hua* School of Computing and Enginring, Univrsity of Huddrsfild, UK J.lu@hud.ac.uk *Institut of Tchnology, Xi'an, Shaanxi, P. R. China Abstract

More information

Energy-Efficient Method to Improve TCP Performance for MANETs

Energy-Efficient Method to Improve TCP Performance for MANETs nrgy-fficint Mthod to Improv TCP Prformanc for MANTs Chaoyu Xiong, Jagol Yim, Jason Ligh and Tadao Murata Computr Scinc Dpartmnt, Univrsity of Illinois at Chicago Chicago, IL 60607, USA ABSTRACT Th currnt

More information

Paper Template and Style Guide for the Vapor Intrusion, Remediation, and Site Closure Conference

Paper Template and Style Guide for the Vapor Intrusion, Remediation, and Site Closure Conference Papr Tmplat and Styl Guid for th Vapor Intrusion, Rmdiation, and Sit Closur Confrnc This Tmplat and Styl Guid dtail th documnt formatting standards and xpctd contnt for a full lngth papr manuscript. Your

More information

JetAdvantage App Handbook

JetAdvantage App Handbook JtAdvantag App Handbook March 2019 HP Inc. maks no warranty of any kind with rgard to this matrial, including, but not limitd to, th implid warrantis of mrchantability and fitnss for a particular purpos.

More information

SPECIFIC CRITERIA FOR THE GENERAL MOTORS GLOBAL TRADING PARTNER LABEL TEMPLATE:

SPECIFIC CRITERIA FOR THE GENERAL MOTORS GLOBAL TRADING PARTNER LABEL TEMPLATE: SPCIFIC CRITRIA FOR TH GNRAL MOTORS GLOBAL TRADING PARTNR LABL TMPLAT: TH TMPLAT IDNTIFIS HOW AND WHR DATA IS TO B PLACD ON TH LABL WHN IT IS RQUIRD AS PART OF A GM BUSINSS RQUIRMNT FONT SIZS AR SPCIFID

More information

Ontology and Context. Isabel Cafezeiro Departamento de Ciência da Computação Universidade Federal Fluminense Niterói - RJ, Brazil

Ontology and Context. Isabel Cafezeiro Departamento de Ciência da Computação Universidade Federal Fluminense Niterói - RJ, Brazil Ontology and Contxt Isabl Cafziro Dpartamnto d Ciência da Computação Univrsidad Fdral Fluminns Nitrói - RJ, Brazil isabl@dcc.ic.uff.br dward Hrmann Hauslr, Alxandr Radmakr Dpartamnto d Informática Pontifícia

More information

RFC Java Class Library (BC-FES-AIT)

RFC Java Class Library (BC-FES-AIT) RFC Java Class Library (BC-FES-AIT) HELP.BCFESDEG Rlas 4.6C SAP AG Copyright Copyright 2001 SAP AG. All Rcht vorbhaltn. Witrgab und Vrvilfältigung disr Publikation odr von Tiln daraus sind, zu wlchm Zwck

More information

I - Pre Board Examination

I - Pre Board Examination Cod No: S-080 () Total Pags: 06 KENDRIYA VIDYALAYA SANGATHAN,GUWHATI REGION I - Pr Board Examination - 04-5 Subjct Informatics Practics (Thory) Class - XII Tim: 3 hours Maximum Marks : 70 Instruction :

More information

Base Schedule Aggregation Portal (BSAP) Interface Specification for BSAP Services

Base Schedule Aggregation Portal (BSAP) Interface Specification for BSAP Services Bas Schdul Aggrgation Portal (BSAP) Intrfac Spcification for BSAP Srvics (Businss Ruls v 9.x(Fall 2017) or latr) Vrsion: 1.3 Dcmbr 19, 2017 Rvision History Dat Vrsion By Dscription 12/19/2017 1.3 WT Additional

More information

Maxwell s unification: From Last Time. Energy of light. Modern Physics. Unusual experimental results. The photoelectric effect

Maxwell s unification: From Last Time. Energy of light. Modern Physics. Unusual experimental results. The photoelectric effect From Last Tim Enrgy and powr in an EM wav Maxwll s unification: 1873 Intimat connction btwn lctricity and magntism Exprimntally vrifid by Hlmholtz and othrs, 1888 Polarization of an EM wav: oscillation

More information

Analysis of Influence AS Path Prepending to the Instability of BGP Routing Protocol.

Analysis of Influence AS Path Prepending to the Instability of BGP Routing Protocol. ISSN : 2355-9365 -Procding of Enginring : Vol.5, No.1 Mart 2018 Pag 1112 Analysis of Influnc AS Path Prpnding to th Instability of BGP Routing Protocol. Hirwandi Agusnam 1, Rndy Munadi 2, Istikmal 3 1,2,3,

More information

Clustering Algorithms

Clustering Algorithms Clustring Algoritms Hirarcical Clustring k -Mans Algoritms CURE Algoritm 1 Mtods of Clustring Hirarcical (Agglomrativ): Initially, ac point in clustr by itslf. Rpatdly combin t two narst clustrs into on.

More information

Impact & Analysis of FHE on nanoscopic TEM Images

Impact & Analysis of FHE on nanoscopic TEM Images Impact & Analysis of FHE on nanoscopic TEM Imags ABSTRACT: TEM imags ar rapidly gaining prominnc in various sctors lik lif scincs, pathology, mdical scinc, smiconductors, fornsics, tc. Hnc, thr is a critical

More information

Reducin} Migratin} secxn:laries

Reducin} Migratin} secxn:laries Rducing Migrating Scondaris Earl Wallr and Ml Parc INLEX, Inc. P.O. Box 1349 Montry, CA. 93942 If you ar lik us, you hav rad a lot rcntly about IMAGE and its 'myths', and you know that thr can b prformanc

More information

FALSE DYNAMIC EIV MODEL IDENTIFICATION IN THE PRESENCE OF NON-PARAMETRIC DYNAMIC UNCERTAINTY

FALSE DYNAMIC EIV MODEL IDENTIFICATION IN THE PRESENCE OF NON-PARAMETRIC DYNAMIC UNCERTAINTY Intrnational Journal of Application or Innovation in Enginring & Managmnt (IJAIEM) Wb Sit: www.ijaim.org Email: ditor@ijaim.org Volum 3, Issu, May 4 ISSN 39-4847 FALSE DYNAMIC EIV MODEL IDENTIFICATION

More information

Architecture of the ATLAS High Level Trigger Event Selection Software

Architecture of the ATLAS High Level Trigger Event Selection Software Architctur of th ATLAS High Lvl Triggr Evnt Slction Softwar S. Armstrong, K.A. Assamagan, J.T. Bains, C.P. B, M. Biglitti, A. Bogarts, V. Boisvrt, M. Bosman, S. Brandt, B. Caron, P. Casado, G. Cataldi,

More information

Dynamic modelling of multi-physical domain system by bond graph approach and its control using flatness based controller with MATLAB Simulink

Dynamic modelling of multi-physical domain system by bond graph approach and its control using flatness based controller with MATLAB Simulink Dnamic modlling of multi-phsical domain sstm b bond graph approach and its control using flatnss basd controllr with MATLAB Simulink Sauma Ranjan Sahoo Rsarch Scholar Robotics Lab Dr. Shital S. Chiddarwar

More information

DO NOW Geometry Regents Lomac Date. due. Similar by Transformation 6.1 J'' J''' J'''

DO NOW Geometry Regents Lomac Date. due. Similar by Transformation 6.1 J'' J''' J''' DO NOW Gomtry Rgnts Lomac 2014-2015 Dat. du. Similar by Transformation 6.1 (DN) Nam th thr rigid transformations and sktch an xampl that illustrats ach on. Nam Pr LO: I can dscrib a similarity transformation,

More information