$B(.(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(B $B(-(B $BF|K\%G!<%?%Y!<%93X2q(B Newsletter $B(-(B 2017$BG/(B12$B7n9f(B ( Vol. 10, No. 7 ) $B(1(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(,(B $B=iE_$N8u!$$O$dG/$N@%$,6a$E$$$F$^$$$j$^$7$?$,!$3'MM$$$+$,$*2a$4$7$G$7$g$&(B $B$+!%G/Kv$4B?K;$N@^$G$O$4$6$$$^$9$,!$$*BN$K$*5$$r$D$1$FNI$-?7G/$r$*7^$($/(B $B$@$5$$(B. $B$5$F!$K\9f$G$O!$(B9$B7n$K3+:E$5$l$^$7$?!$%f%S%-%?%9%3%s%T%e!<%F%#%s%0J,Ln$N%H(B $B%C%W%+%s%U%!%l%s%9!V(BUbiComp$B!W$H%&%'%"%i%V%k%3%s%T%e!<%F%#%s%0J,Ln$N%H%C%W(B $B%+%s%U%!%l%s%9!V(BISWC$B!W$r$O$8$a!$>pJs8!:w!$%G!<%?%Y!<%9!$%J%l%C%8%^%M%8%a(B $B%s%H$*$h$S%^%k%A%a%G%#%"J,Ln$G$N:G=EMW9q:]2q5D!V(BCIKM$B!W$H!V(BACM Multimedia$B!W(B $B$K$D$$$F$44s9F$$$?$@$-$^$7$?!%$=$l$>$l$N2q5D$NFCD'$d:G6a$N798~!$%H%C%W%+(B $B%s%U%!%l%s%9$X$NEj9F$N%a%j%C%H!$O@J8:NBr$K;j$k$^$G$N9)IW$J$I!$3'MM$N$4;2(B $B9M$K$J$l$P9,$$$G$9!%2C$($F!$(B9$B7n$K3+:E$5$l$^$7$?!$(BDB$B7O!&(BWeb$B7O$K$*$1$k9qFb(B $BM#0l$N::FIIU$-2q5D$G$"$k(BWebDB Forum 2017$B$N3+:EJs9p$r$44s9F$$$?$@$-$^$7$?!%(B $BK\9f$J$i$S$K(BDBSJ Newsletter$B$KBP$9$k$40U8+$"$k$$$O<!9f0J9_$K4|BT$9$kFbMF(B $B$K$D$$$F$N$40U8+$,$4$6$$$^$7$?$i(Bnews-com [at] dbsj.org$B$^$G$*4s$;$/$@$5$$!%(B $BF|K\%G!<%?%Y!<%93X2q(B $BEE;R9-JsJT=80Q0w2q(B $B!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!JC4EvJT=80Q0w!!2&(B $B8585!K(B $B!!!!(B ======================================================================= ---- $BL\<!(B ---- $B#1(B. UbiComp/ISWC 2017 $B;22CJs9p(B $B>>0f=Y(B $B?@8MBg3X(B $B#2(B. UbiComp 2017 $B;22CJs9p(B $B?\%vyu@;?M(B $BEl5~9)6HBg3X(B $B#3(B. CIKM 2017 $B;22CJs9p(B $BM{5HV((B $B5~ETBg3X(B $B#4(B. ACM Multimedia 2017 $B;22CJs9p(B $BF!%F%$%F%$(B $BF|K\EE5$3t<02q<R(B $B#5(B. WebDB Forum 2017 $B3+:EJs9p(B $BEOJUCN7CH~(B $B;:6H5;=QBg3X1!Bg3X(B ======================================================================= $B"##1"#(B UbiComp/ISWC 2017 $B;22CJs9p!!!!!!!!!!!!(B $B!!!!!!!!>>0f=Y(B ($B?@8MBg3X(B) $B!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!(B $B!!!!!!!!!!!!!!!!!!!!(B 2017$BG/(B9$B7n(B11$BF|!A(B15$BF|$K3+:E$5$l$?!V(BUbiComp/ISWC 2017$B!W$K;22CCW$7$^$7$?!%(B $B3+:ECO$O%O%o%$$N%^%&%$Eg$K$"$k(BGrand Wailea$B$H$$$&%[%F%k$G$9!%(BUbiComp$B$O%f(B $B%S%-%?%9%3%s%T%e!<%F%#%s%0J,Ln!$(BISWC$B$O%&%'%"%i%V%k%3%s%T%e!<%F%#%s%0J,(B $BLn$N%H%C%W%+%s%U%!%l%s%9$G!$9g7W(B561$BL>$,;22C$7$F$*$j$^$7$?!%(BUbiComp$B$G$O!$(B $B:#G/$O(B576$B7o$NEj9F$+$i(B74$B7o$,:NBr$5$l:NBrN($,(B20.8$B!s$H$J$j!$(BISWC$B$G$O(B90$B7o$N(B $BEj9FCf(Bbrief (poster)$B$r4^$a(B34$B7o$,:NBr$5$l!$:NBrN($,(B32.1%$B!J(Blong$B!\(Bshort$B!K(B $B$H$J$C$F$*$j$^$7$?!%(BISWC$B$O:rG/$N:NBrN($KHf$Y$k$H>e$,$j$^$7$?$,!$$I$A$i(B $B$bAjJQ$o$i$:9b$$6%AhN($G$7$?!%(B $B2q5D$N%9%1%8%e!<%k$H$7$F$O!$(B9$B7n(B11$B!$(B12$BF|$K%o!<%/%7%g%C%W$,$"$j(B13$BF|!A(B15$BF|(B $B$KK\2q5D$,9T$o$l$^$7$?!%%o!<%/%7%g%C%W$O(B10$B8D$"$j%f%S%-%?%9$d%&%'%"%i%V(B $B%k$K4XO"$9$k$b$N$G$7$?!%K\2q5D$N%W%m%0%i%`$O!$(BUbiComp$B$H(BISWC$B9g$o$;$F(B37$B%;(B $B%C%7%g%s$G9=@.$5$l$F$$$^$7$?!%;d$N8&5f$G$"$k!V(BSmart Eye Mask: Sleep Sensing System Using Infrared Sensors$B!W$O(BISWC$B$G:NBr$5$l!$(BWEARABLES 1$B$H(B $B$$$&%;%C%7%g%s$GH/I=CW$7$^$7$?!%$3$NO@J8$O!$%"%$%^%9%/7?%G%P%$%9$r:n@=(B $B$7!$?gL2>uBV!J(BREM$B?gL2$+(BNREM$B?gL2!K$rG'<1$9$k$H$$$&8&5f$G$9!%?gL2>uBV$N>p(B $BJs$+$i?gL2>c32$N<#NE$KLrN)$C$?$j!$5$;}$A$N$h$$L\3P$a$,2DG=$K$J$j$^$9!%(B $B?gL27WB,%G%P%$%9$r4JC1$KAuCe$G$-!$$I$s$J4D6-2<$G$bB,Dj$G$-$k$H$$$&ItJ,(B $B$,$3$N8&5f$N%]%$%s%H$G$9!%::FI$G$N%l%S%e!<$G$O!$8&5f$N%"%$%G%"$rFC$KI>(B $B2A$7$F$$$?$@$-!$:NBr$K;j$j$^$7$?!%8&5f$N@.2L$r@lLgJ,Ln$N%H%C%W%+%s%U%!(B $B%l%s%9$GH/I=$9$k$3$H$,$G$-!$$^$?$=$3$GMM!9$JD0=0$+$iM-0U5A$J%U%#!<%I%P(B $B%C%/$r$$$?$@$/$3$H$,$G$-!$$H$F$b5.=E$JBN83$r$5$;$F$$$?$@$-$^$7$?!%B>$N(B $B?M$NH/I=$G$O!$?75,@-$N$"$kLLGr$$%"%$%G%"$d%"%W%m!<%A$N8&5f$,H/I=$5$l$F(B $B$*$j!$BgJQJY6/$K$J$j$^$7$?!%$^$?%f%S%-%?%9!$%&%'%"%i%V%k$NJ,Ln$NH/I=Cf(B $B$K%G%b$r9T$&J}$bB?$/!$2q>l$NJ70O5$$O@9$j>e$,$C$F$$$^$7$?!%(B $B2q>l$N%O%o%$$N%^%&%$Eg$OH~$7$$%S!<%A$K7C$^$l!$9b5i%[%F%k$G$N%j%>!<%H%i(B $B%$%U$d%7%g%C%T%s%0$b3Z$7$a$kEg$G$9!%2HB2N99T$d%O%M%`!<%s$G$$$i$C$7$c$k(B $BJ}$bB?$$$i$7$/!$3Z$7$$M[5$$JJ70O5$$KJq$^$l$F$$$^$7$?!%$^$?!$%W%m%0%i%`(B $B$K$O%S!<%A%?%$%`$,@_$1$i$l$F$*$j!$%O%o%$$N3$$r3Z$7$`$3$H$,$G$-$^$7$?!%(B $BMhG/$K3+:E$5$l$k(BUbiComp/ISWC 2018$B$O(B10$B7n$K%7%s%,%]!<%k$N(BSuntec Convention Center$B$G9T$o$l$kM=Dj$G$9!%@'Hs$H$bEj9F!$;22C$r$48!F$$/$@$5$$!%(B ($B>>0f=Y(B $B?@8MBg3X(B $BBg3X1!9)3X8&5f2J(B $BDMK\!&;{ED8&5f<<(B) ----------------------------------------------------------------------- $B"##2"#(B UbiComp 2017 $B;22CJs9p(B $B!!!!!!(B $B?\%vyu@;?M(B ($BEl5~9)6HBg3X(B) UbiComp2017$B2q5D$,(B2017$BG/(B9$B7n(B11$BF|$+$i(B15$BF|$N4V$K!$%"%a%j%+!&%O%o%$$G3+:E$5(B $B$l$^$7$?!%$3$A$i$N2q5D$N%a%$%s%F!<%^$OL>A0$NDL$j(BUbiquitous Computing$B$H(B $B$J$C$F$$$^$9!%$3$l$O!$<R2q$H%3%s%T%e!<%?$N6&B8!&JP:_2=$5$;$k$H$$$&$3$H(B $B$rL\I8$K!$%S%C%/%G!<%?$d%&%'%"%i%V%kC<Kv!$%9%^!<%HC<Kv$rMx3hMQ$7$?5;=Q(B $B$K4X$9$kJ,Ln$G$9!%:#G/$+$i!$(BInteractive, Mobile, Wearable and Ubiquitous Technologies (IMWUT) $B$H$$$&%8%c!<%J%k$K:NBr$5$l$?$b$N$,2q5D$GH/I=$G$-$k(B $B$H$$$&7ABV$K$+$o$j!$:#2sH/I=$NBP>]$H$J$kHO0O$N3X2q;o$NEj9F?t$O(B576$BK\!$:N(B $BBr?t$O(B122$BK\!$:NBrN($O$*$*$h$=(B2$B3d$G$"$j$^$7$?!%(BISWC$B$H$$$&2q5D$HJBNs$G9T(B $B$o$l!$2q5D$O(B5$B$D$N%Q%i%l%k%;%C%7%g%s$G?J9T$7!$;22C<T$O(B561$B?M!JF|K\?M$O(B78 $B?M!K$G$7$?!%(B $B2q5D$NFbMF$O!$?M4V$N9TF0!J9TF0%m%0$d%3%s%F%-%9%H!K$d7r9/!J1?F0Jd=u$+$i(B $BIaCJ$N7l05Ey$N7r9/;XI8%b%K%?%j%s%0!$@:?@<@45$^$G!K!$ET;T2r@O$+$i!$$=$l(B $B$i$N5;=Q$K;HMQ$5$l$k4pAC5;=Q$^$GMM!9$JJ,Ln$KEO$j$^$9!%$=$N$h$&$J9-$$HO(B $B0O$N8&5f$,=8$^$k2q5D$G$9$,!$$=$l$i$K6&DL$9$kItJ,$H$7$F$O!$%G!<%?$NMx3h(B $BMQ$d!$%&%'%"%i%V%kC<Kv$d%9%^!<%HC<Kv!$4D6-Fb%;%s%5$r3hMQ$7$J$,$i2r7h$9(B $B$k$H$$$&ItJ,$G$9$G$9!%(B $B;d$O(BUbiComp$B=i;22C$G$7$?$,!$2q5D$K;22C$9$k$3$H$G<+J,$N8&5f%F!<%^$NOC0J30(B $B$K$bMM!9$JJ,Ln$N>pJs$,F@$i$l!$$H$F$b?7$7$$CeA[$X$N=u$1$K$J$k2q5D$G$"$k(B $B$H46$8$^$7$?!%FC$K!$:#2s$O5!3#3X=,$r<h$jF~$l$??7$7$$%"%W%j%1!<%7%g%s$J(B $B$I$bB?$/!$$H$F$bJ9$-$4$?$($N$"$kFbMF$G$7$?!%(B $B;d$O$3$A$i$N2q5D$K0J2<$N:NBr$5$l!$8}F,H/I=$G;22C$5$;$F$$$?$@$-$^$7$?!%(B Masato Sugasaki and Masamichi Shimosaka. Robust Indoor Localization across Smartphone Models with Ellipsoid Features from Multiple RSSIs. Proc. ACM IMWUT. 1, 3, Article 103 (September 2017), 16 pages. $B$3$A$i$NO@J8$NFbMF$r4JC1$K@bL@$9$k$H!$(BWi-Fi$B$N?.9f6/EY$rMQ$$$k20FbB,0L5;(B $B=Q$K$*$$$F!$MM!9$J<oN`$NC<Kv$GE}0lE*$K9b@:EY$JB,0L$r9T$($k$h$&$J<jK!$r(B $B9=C[$7$?$b$N$H$J$C$F$$$^$9!%(B $B$3$N2q5D$O!$=i$a$F$N8}F,H/I=$G$7$?!%$3$N?7$?$J7P83$G0lHV46$8$?$3$H$O!$(B $B@$3&$N%H%C%W$N8&5f<T$K(Breview$B$d!$<ALd$r$$$?$@$/$3$H$G!$8&5f$K$*$1$kF;I8(B $B$d<!$N%9%F%C%W$NBg$-$J=u$1$K$J$k$H$$$&$3$H$G$9!%%H%C%W$N2q5D$KD)@o$7$F(B $B$$$/$3$H$O!$8&5f$r$h$j$h$$>l=j$GH/I=$9$k$@$1$G$J$/!$8&5f$N<A$r9b$a$k$?(B $B$a$K$b=EMW$J$3$H$G$"$k$H46$8$^$7$?!%;d$O=$;N2]Dx(B2$BG/$H8&5f<T$H$7$FJb$_;O(B $B$a$?$P$+$j$G$9$,!$@'Hs$3$l$+$i$bD)@o$7B3$1$F$$$3$&$H;W$$$^$9!%;d$O<c<j(B $B$NCf$G$b<c<j8&5f<T$G62=L$G$9$,!$@'Hs3'$5$s$b@$3&$KD)@o$7!$8&5f$N<A$r9b(B $B$a$F$$$-$^$7$g$&!%(B ($B?\%vyu@;?M(B $BEl5~9)6HBg3X(B $BBg3X1!>pJsM}9)3X8&5f2J(B $B2<:d8&5f<<(B) ----------------------------------------------------------------------- $B"##3"#(B CIKM 2017 $B;22CJs9p(B $B!!!!!!!!!!!!!!!!!!!!!!!!!!!!M{5HV((B ($B5~ETBg3X(B) The 26th ACM International Conference on Information and Knowledge Management (CIKM2017) has just been held in Singapore at November 6-10, 2017. I am very glad to be able to attend this high-level international conference. Instead of introducing the detailed papers related to my research area, i.e., crowdsourcing and human computation, I would like to introduce some general issues, which may be interested by more readers of this report. 1. Conference and Singapore In contrast to the conferences with researchers from single community, CIKM contains attendances from diverse communities including ML/DM, IR, DB/DE, NLP and so on. As claimed by the organizers, this conference is growing on both the size and topic diversity, and will continue to grow in the near future. There are more than 700 attendees from more than 40 countries this year. The total number of submission of full papers is 855 (701 in 2016 and 646 in 2015). The acceptance rate is 171/855 (20%) for full papers and 119/419 (28%) for short papers. The theme of the conference this year is smart cities and smart nations. In the keyword cloud of the papers, graph/network, entity, semantic and recommendation are some of the major topics. CIKM2018 will be held in Turin, Italy. CIKM2019 will be held in Beijing, China. For the impression of Singapore, both the foods provided by the conference and in the streets are very delicious. In this country, it is very easy to find the foods from various Asian countries. Singapore is mainly a financial center, while there are IT companies located at Singapore which focus on the southeast Asian market. 2. Keynotes - The talk given by Dr. Rajeev Rastogi from Amazon introduced the machine $B!!(Blearning research and applications in Amazon, including product demand $B!!(Bforecasting, product classification, product matching, information $B!!(Bextraction from reviews, question answering system and so on. In addition, one of interesting topics was to standardize the diverse shoe sizes across brands and locales. These research topics related to practical requirements are very interesting and useful. - Dr. Qiang Yang from Hong Kong University of Science and Technology gave a talk on deep transfer learning, which adapted the model trained from the data from a task/domain to other data from other tasks/domains. It is one of the hot topics in recent years for various AI research areas. He is one of the most famous researchers on this topic. - Dr. Rada Mihalcea from University of Michigan talked about deception $B!!(Bdetection, which tried to detect whether a person lied or told the $B!!(Btruth. In the early years, they started this research only based on $B!!(Btext; after that they considered the language from different culture; $B!!(Bthen they focused on multi model information including motion and facial expression in the video as well as text. Their methods showed better performance than human beings on deception detection. They also considered language demographics and fake news in recent work. 3. Best Paper Award Y. Zhuang et al "Hike: A Hybrid Human-Machine Method for Entity Alignment in Large-Scale Knowledge Bases". The authors are from Tsinghua unviversity, UC Irvine and the University of Hong Kong. It is to align the entities when integrating the heterogeneous Knowledge Bases (KBs). Due to the inconsistency and uncertainty of large-scale KBs, automatic techniques for KBs alignment achieve low quality. This paper thus proposes a hybrid human-machine framework by harnessing the crowd to improve the alignment quality. It is a solid work on such type of crowdsourcing and human computation topics. It proposes a rational solution which covers the required issues in such topics including crowdsourcing framework, cost optimization and quality control. At the beginning, I thought it was a similar work with "Reducing Uncertainty of Schema Matching via Crowdsourcing" in VLDB 2013 (not referred by this paper). This VLDB2013 paper also proposed a hybrid method and asked crowd to match the elements from two schemata (e.g., database schemata, XML schemata). However, because the entities in knowledge graph are different the elements in schemata. The authors of this CIKM2017 paper propose a different framework and focus on specific issues for this entity and knowledge graph scenario such as entity partition in knowledge graph to reduce the cost of crowd. 4. AnalytiCup 2017 It contains four data competitions which lasts nine months in total. They are Shenzhen Meteorological Bureau-Alibaba Short-Time Quantititative Precipitation Forecasting Challenge; DataSpark Mobility Open-Task Challenge; Lazada Product Title Quality Challenge; DHL Temperature-Controlled Supply Chain Hackathon Challenge. There are 1500 teams from all over the world. Specially, the Dataspark company provides some useful mobility data of Singapore, e.g., "footfall" is the unique count of people appearing at a given place and time; "origin-destination" is the unique count of people traveling from a given origin to a given destination area; "dwell time" is the average time spent by visitors who entered a given location at a given hour or day. These data are not easy to obtain for the academic researchers in the universities and very useful for some potential research topics. This is a brief introduction of the programs in the CIKM 2017. If you are interested in any issues and have questions, please feel free to contact me. ($BM{5HV((B $B5~ETBg3X(B $BBg3X1!>pJs3X8&5f2J(B $BFCDj8&5f0w(B) ----------------------------------------------------------------------- $B"##4"#(B ACM Multimedia 2017 $B;22CJs9p(B $B!!(B $BF!%F%$%F%$(B ($BF|K\EE5$3t<02q<R(B) 2017$BG/(B10$B7n(B23$BF|$+$i(B27$BF|$K$+$1$F3+:E$5$l$?(BACM Multimedia 2017$B!J(BMM 2017$B!K(B $B$K;22C$7!$%G%bH/I=$r9T$C$?!%:#2s$N3+:ECO$O%7%j%3%s%P%l!<$K$"$k(BMountain View$B$G!$(BGoogle$B$d(BFacebook$B$J$IB??t$N(BIT$B4k6H$H%9%?%s%U%)!<%IBg3X$J$I$NBg3X(B $B!&8&5f5!4X$,=8$^$C$F$$$k!%;22C<T?t$O(B700$B?M$K>e$j!$2q4|Cf$KM-L>(BIT$B4k6H$a$0(B $B$j$H%9%?%s%U%)!<%IK,Ld$O?M5$$r=8$a$?!%2q>l$O@$3&:GBg5i$N=EMW$J%3%l%/%7(B $B%g%s$r=jM-$7$F$$$k%3%s%T%e!<%?Nr;KGnJ*4[$G!$%3%s%T%e!<%?Nr;K>e$G0UL#?<(B $B$$%^%7%s$d@=IJ$N%b%G%k$J$I$,E8<($5$l$F$$$k!%:#$^$G(BDatabase$B$K4X$o$C$F$$(B $B$?;d$K$H$C$F$O!$=i$a$F$N%^%k%A%a%G%#%"J,Ln$N3X2q$@$C$?!%(B ACM Multimedia$B$O%^%k%A%a%G%#%"J,Ln$K$*$1$k%H%C%W9q:]2q5D$G!$:#G/$O(B25$B<~(B $BG/$r7^$($k!%:#G/$+$i=i$a$F!$(BLong paper$B$H(BShort paper$B$rF1$8Ej9F%H%i%C%/$H(B $B::FI%W%m%;%9$KE}0l$7$?!%=>Mh$NO@J8D9$5!J(BShort paper: 5$B%Z!<%8(B, Long paper: 8$B%Z!<%8(B+$B0zMQ!K$+$i!$=@Fp$JHO0O!J(B6-8$B%Z!<%8(B+$B0zMQ!K$KJQ99$7!$$h$j0l4S@-$,(B $B$"$k::FI%W%m%;%9$rL\;X$9!%K\2q5D$NO@J8Ej9F?t$O(B684$B7o$G!$:NBrN($O(B27.63%$B$@(B $B$C$?!%$=$N$&$A!$8}F,$O(B7.15%$B!$%]%9%?!<$O(B20.74%$B!$(BBest paper$B$O(B0.58%$B$G!$6%(B $BAh$,$9$4$/7c$7$+$C$?!%(B22$B$N4X?4NN0h$,!$=>Mh$N(B4$B%F!<%^!J(BSystem and Application, Experience, Understanding, Engagement$B!K$K2C$($F!$?7%F!<%^$H$7$F(BNovel Topics$B$,F3F~$5$l!$(B5$B$N%F!<%^$K%0%k!<%W$5$l$?!%(B $B:#G/$G$O!$(BStarkey$B$d(BNVidia$B!$(BSamsung$B!$(BHTC$B!$(BGoogle$B!$(BUnity$B$+$i!$3JJL$K(B6$B7o$N(B Keynote$B9V1i$,$"$j!$K\2q5D$N;0F|4V$G$O!$KhF|8aA0(B1$B7o8a8e(B1$B7o$N$[$I!$=<<B$7(B $B$?%W%m%0%i%`$@$C$?!%%H%T%C%/%9$O(BAI$B$K$h$k?M4VCN3P$N6/2=$H%X%k%9%1%"$N8~(B $B>e$+$i!$(BDeep learning$B$K$h$k%O!<%I%&%'%"?J2=$d!$%9%^!<%H%U%)%s$N%$%s%?!<(B $B%U%'!<%9$N%^%k%A%b!<%@%k2=$X$HB?4t$K$o$?$k!%(BKeynote$B9V1i$N$[$+!$%A%e!<%H(B $B%j%"%k(B6$B7o!$%Q%M%k(B2$B7o!$%G%b(B23$B7o!$%S%G%*(B1$B7o!$(BBrave New Idea 4$B7o!$(BGrand Challenge 10$B7o!$(BBusiness Idea Venture 2$B7o!$(BOpen Source Competition 6$B7o!$(B Interactive Art Presentation 8$B7o!$(BThematic Workshop 4$B7o!$(BWorkshop 10$B7o(B $B$,$"$j!$%W%m%0%i%`$,B?<oB?MM$G!$K-IY$@$C$?!%(B $B:#G/$N(BBest paper$B$OCf9qEE;R2J5;Bg3X$+$i(BHeng Tao Shen$B65<x$N%0%k!<%W$N(B $B!V(BAdversarial Cross-Modal Retrieval$B!W$@$C$?!%%F%-%9%H$d;k3P!$2;@<$H$$$C(B $B$?%b!<%@%k$r2#CG$7$?>pJsCj=P$KCmL\$7!$%b!<%@%k$K0MB8$7$J$$6&DLItJ,6u4V(B $B$r3X=,$9$kJ}K!$,Ds0F$5$l$?!%$[$+$K$b!$%^%k%A%b!<%@%k!J(BMulti-modal$B!K$d%b(B $B!<%@%k2#CG!J(BCross-modal$B!K!$%a%G%#%"2#CG!J(BCross-media$B!K$J$I$K4X$9$kH/I=(B $B$,B??t$"$j!$%^%k%A%a%G%#%"J,Ln$N%H%l%s%I%H%T%C%/$H$J$C$F$$$k!%%Q%M%k$G(B $B$b$3$N%H%T%C%/$,<h$j>e$2$i$l$?!%(B $B;d$N%G%bH/I=!V(BDiversified and Summarized Video Search System$B!W$G$O!$1G(B $BA|8!:w$K$*$1$k7k2L=EJ#$N2]Bj$r2r7h$9$k!%4i$dG/Np!$@-JL!$I~$N?'$J$IJ#?t(B $B$NFCD'NL$rMQ$$!$N`;wKt$O=EJ#$7$F$$$k7k2L$r$^$H$a$F!$B?MM2=$7$?BeI=E*$J(B $B7k2L$rDs<($9$k$3$H$,$G$-$k!%$=$l$K$h$j!$F10l?MJ*$,;}$DJ#?t$NFCD'$r$^$H(B $B$a$k$3$H$,$G$-!$MF5?<T$HLB;R$NA\::$d5RAX$NJ,@O$J$I$KLr$KN)$D$3$H$,9M$((B $B$i$l$k!%:rG/F~<R$7$F$+$i!$1GA|8!:w$K4X$9$k8&5f$r;O$a!$0lG/4VH>?'!96lO+(B $B$7$?$,!$3X2qH/I=$K;22C$G$-@=IJ2=$K$b4|BT$G$-$F!$$H$F$b8&5f<T$H$7$F$N$d(B $B$j$,$$$,46$8$i$l!$$&$l$7$+$C$?!%(B $BMhG/$N(BMM$B$O(B2018$BG/(B10$B7n(B22$BF|$+$i(B26$BF|!$4Z9q$N%=%&%k$G3+:E$5$l$kM=Dj!%KLD+A/(B $B$NLdBj$G0BA4$r?4G[$9$kJ}$,B?$$$H;W$o$l$k$,!$>u67$K1~$8$F9a9A$XJQ99$9$k(B $BBP:v$,$"$k$HJ9$$$?!%%^%k%A%a%G%#%"J,Ln$K6=L#$r;}$?$l$?J}$O!$@'Hs$H$bEj(B $B9F!$;22C$r$48!F$$/$@$5$$!%(B ($BF!%F%$%F%$(B $BF|K\EE5$3t<02q<R%7%9%F%`%W%i%C%H%U%)!<%`8&5f=j(B) ----------------------------------------------------------------------- $B"##5"#(B WebDB Forum 2017 $B3+:EJs9p(B $BEOJUCN7CH~(B ($B;:6H5;=QBg3X1!Bg3X(B) $B!!!!!!!!!!!!!!3t<02q<R(BFRONTEO $B!!!!!!!!!!!!!!3ZE73t<02q<R!!3ZE75;=Q8&5f=j(B $B!!!!!!!!!!!!!!(BSansan$B3t<02q<R(B $B!!!!!!!!!!!!!!3t<02q<RF|N)@=:n=j(B $B!!!!!!!!!!!!!!3t<02q<REl<G(B $B!!!!!!!!!!!!!!IY;NDL3t<02q<R(B $B!!!!!!!!!!!!!!3t<02q<RGnJsF2(B $B!!!!!!!!!!!!!!(BGoogle $B!J%7%k%P!<!K!'3t<02q<R%5%$%P!<%(!<%8%'%s%H(B $B!!!!!!!!!!!!!!F|K\%"%$!&%S!<!&%(%`3t<02q<R!!El5~4pAC8&5f=j(B $B!!!!!!!!!!!!!!3t<02q<R%[%C%H%j%s%/(B $B!!!!!!!!!!!!!!3t<02q<R(BGunosy $B!J%V%m%s%:!K!'IY;N%<%m%C%/%93t<02q<R(B $B!!!!3t<02q<R%S%k%G%#%C%H(B $B!!!!!!!!!!!!!!(BSRA OSS, Inc. $BF|K\;Y<R(B $B!{6(;?!'>pJs=hM}3X2q(B $B>pJs4pAC$H%"%/%;%95;=Q8&5f2q(B $B!!!!!!!!(BARG Web$B%$%s%F%j%8%'%s%9$H%$%s%?%i%/%7%g%s8&5f2q(B $B!!!!!!!!(BACM SIGMOD$BF|K\;YIt(B $B!!!!!!!!C^GHBg3XCNE*%3%_%e%K%F%#4pHW8&5f%;%s%?!<(B $B!!!!!!!!(BFirebird$BF|K\%f!<%62q(B $B!!!!!!!!F|K\(BPostgreSQL$B%f!<%62q(B $B!!!!!!!!F|K\(BMySQL$B%f!<%62q(B WebDB Forum 2017$B$O!$(B2017$BG/(B9$B7n(B18$BF|$+$i(B20$BF|$K$+$1$F$*Cc$N?e=w;RBg3X$K$F(B $B3+:E$5$l$^$7$?!%K\%U%)!<%i%`$O!$(BWebDB$B%3%_%e%K%F%#$K$*$1$k;:3XO"7H$r<4(B $B$K?x$($F$$$k$3$H$H!$<A$N9b$$8&5fH/I=$N>l$G$"$k$3$H$,Bg$-$JFCD'$G$9!%(B $B:#2s$O(B2008$BG/$K(BWebDB Forum$B$HL>$rJQ$($F$+$i(B10$B2sL\$N5-G0$9$Y$-2q$H$J$j$^(B $B$7$?!%;22C<T?t$O(B300$BL>$rD6$(!$>7BT9V1i$dFCJL%;%C%7%g%s!$0lHLH/I=(B, $B%9%](B $B%s%5!<4k6H$K$h$k:G?75;=QH/I=!J%F%/%N%m%8!<%7%g!<%1!<%9!K(B, $B4k6H%9%]%s%5(B $B!<$K$h$kCk?)2q(B, $BE8<(%V!<%9$J$IB?:L$J%$%Y%s%H$GBg$$$K@9$j>e$,$j$^$7$?!%(B WebDB Forum 2017$B$O!VD6%9%^!<%H<R2q$N%(%s%8%s$r%G%6%$%s$9$k(B#webdbf2017$B!W(B $B$r%F!<%^$H$7!$B?:L$J>7BT9V1i$dFCJL%;%C%7%g%s$r4^$a!$0J2<$NH/I=$,9T$o$l(B $B$^$7$?!%(B $B!&>7BT9V1i!'CO0h(BIoT$B$H>pJsNO$K$h$k<R2q$N%P!<%8%g%s%"%C%W(B $B!&(BiDB$B>7BT9V1i!'(BLearning from Bandit Feedback $B!&FCJL%;%C%7%g%s!'(BFinTech$B$N:G?7F08~$H:#8e$NE83+(B $B!&(BOSS$BFCJL%;%C%7%g%s!'%*!<%W%s%=!<%9%3%_%e%K%F%#$N3hF0$H8&5f<T!&3X@8$X(B $B$N4|BT(B $B!&0lHLO@J8%;%C%7%g%s(B(53$B7o(B) $B!!(BTOD$B0lHLO@J8(B12$B7o!$(BTOD$B%F%/%K%+%k%N!<%H(B8$B7o!$(BDBS/IFAT$B9gF18&5f2q(B31$B7o!$(B $B!!!!(BDE$B8&5f2q(B11$B7o(B $B!&%F%/%N%m%8!<%7%g!<%1!<%9!'(B21$B7o!J$&$A%i%$%H%K%s%0%H!<%/(B6$B7o!K(B $B!&3$30$NCxL>$J8&5f<T$r%3%a%s%F!<%?$K7^$($FGn;N3X@8$N%a%s%?%j%s%0$r9T$&(B $B!!(BiDB$B%_%K%o!<%/%7%g%C%W(B(4$B7o(B) $B!&%]%9%?!<H/I=(B 65$B7o!J0lHLH/I=(B42$B7o(B, $B4k6H(B21$B7o(B, $B6(;?(B2$B7o!K(B $B!&;22C<T4V$N8rN.$H5;=QF$O@$N$?$a$N%]%9%?!<%l%;%W%7%g%s(B $B:rG/EY$+$iEj9F7ABV$,JQ$o$j!$(BTOD$B$N0lHLO@J8$*$h$S%F%/%K%+%k%N!<%H$H$7$F(B $B:NO?$5$l$?O@J8$,(BWebDB Forum$B$K$*$$$FH/I=$G$-$k$h$&$K$J$j$^$7$?!%$^$?>p(B $BJs=hM}3X2q(BDBS$B8&5f2q!&(BIFAT$B8&5f2q$*$h$SEE;R>pJsDL?.3X2q(BDE$B8&$H$N9gF13+:E(B $B$H$J$j!$40@.EY$N9b$$O@J8$+$iK(2jE*$J$b$N$^$GI}9-$$%9%F!<%8$r%+%P!<$9$k(B $B$h$&$K$J$j$^$7$?!%Bg5,LO%G!<%?$KBP$9$k%9%1!<%i%V%k$J=hM}(B, $B9bB.%9%H%l!<(B $B%8%G%P%$%9$KBP1~$7$?%G!<%?%9%H%"9=B$$J$I$N%G!<%?%Y!<%94pHW5;=Q$K2C$(!$(B $BCOM}>pJs%G!<%?$d%0%i%U%G!<%?$J$I$NB?MM$J%G!<%?$KBP$9$kJ,@O4pHW5;=Q!$?<(B $BAX3X=,Ey$rMxMQ$7$?5!3#3X=,%b%G%k$NDs0F!&1~MQ$J$I$N8&5fH/I=$J$I!$(BWebDB Forum$B$i$7$$B?:L$JH/I=$,9T$o$l$^$7$?!%(B $B$5$i$K!$:#G/$O%9%]%s%5!<4k6H$K$h$k5;=QJs9p%;%C%7%g%s$r%F%/%N%m%8!<%7%g(B $B!<%1!<%9$H$$$&L>A0$K0l?7$7$^$7$?!%(B20$B<R$N%9%]%s%5!<4k6H$H(B7$BCDBN$K$46(;?!&(B $B$48e1g$$$?$@$-!$5;=QJs9pH/I=$d>o@_E8<(!$:)?F2q$dCk?)2q$J$IB?$/$N%$%Y%s(B $B%H$GHs>o$K@9$j>e$,$j$^$7$?!%>o@_E8<(%3!<%J!<$O%a%$%s2q>lA0$N%*!<%W%s%9(B $B%Z!<%9$K$"$C$F!$3+:E=iF|$NBfIw0l2a$K$h$k5(@a30$l$N5$29>e>:$b$"$j>o@_$9(B $B$k$K$O>/!9%O!<%I$J>u67$K$J$j$^$7$?$,!$Cc2[%3!<%J!<J;@_$G5Y7F;~4V$K$O;2(B $B2C<T$,%3!<%R!<$d$*2[;R$rJR<j$KOB$d$+$K5DO@$9$k$H$H$b$KM-8z$J%M%C%H%o!<(B $B%-%s%0$N>l$H$J$j$^$7$?!%(B $B$^$?:#2s$O=i$a$FBw;y%5!<%S%9$rF3F~$7$^$7$?!%3X2qCf$N;R6!$NMB$1@h$,$J$/(B $B1s3VCO$X$N3X2q;22C$O$J$+$J$+Fq$7$$$HD|$a$F$$$?J}!9$K$b0B?4$7$F$4;22C$$(B $B$?$@$/$N$KI,MW$J%5!<%S%9$+$H;W$$$^$9!%;d<+?H$b:#2s(B1$B:P$N2f$,;R$rMB$1$^(B $B$7$?!%2f$,;R$OM%$7$$J]0i;N$NJ}$H(B3$BF|4V3Z$7$/2a$4$7;d$b%U%)!<%i%`;22C$K(B $B=8Cf$G$-$F%[%C$H$7$^$7$?!%:#8e$b0z$-B3$-F3F~$5$l$k$3$H$r4j$C$F$$$^$9!%(B $BMM!9$J<u>^$N5!2q$,$"$k$@$1$G$J$/!$B?$/$N8=Lr$N4k6H8&5f<T$HD>@\CN$j9g$$!$(B $B>pJs8r49$G$-$k5.=E$J5!2q$G$"$k(BWebDB$B%U%)!<%i%`!%:#8e$b3X@8$N3'MM$+$i$N(B $B@Q6KE*$JEj9F$r4|BT$7$F$*$j$^$9!%(B $B:G8e$K$J$j$^$7$?$,!$$46(;?D:$$$?4k6H5Z$SCDBN$N3'MM!$$4EPCE!&$4;22CD:$$(B $B$?3'MM$K8|$/$*Ni?=$7>e$2$^$9!%(B ($BEOJUCN7CH~(B $B;:6H5;=QBg3X1!Bg3X(B $B=Z65<x(B) ======================================================================= -- $B2&(B $B8585(B $B;38}Bg3XBg3X1!AO@.2J3X8&5f2J(B $B9)3X7O3X0h!&CNG=>pJs9)3XJ,Ln(B $B!J7sC4!'9)3XItCNG=>pJs9)3X2J!K(B $B=u65!!Gn;N!J4D6-?M4V3X!K(B y.wang@yamaguchi-u.ac.jp circl.wang@gmail.com Yuanyuan Wang, Ph.D Assistant Professor, Information Science and Engineering Section, Department of Engineering, Graduate School of Sciences and Technology for Innovation, Yamaguchi University, Japan$B!!(B E-mail: y.wang@yamaguchi-u.ac.jp circl.wang@gmail.com