暂无图片
暂无图片
暂无图片
暂无图片
暂无图片
基于自注意力网络的共享账户跨域序列推荐 - 郭磊, 李秋菊, 刘方爱, 王新华.pdf
115
14页
0次
2021-11-10
免费下载
DOI
:
issn
JournalofCom
p
uterResearchandDevelo
p
ment
(
):
,
 
稿
:
;
:
 
:
(
,
);
(
M
)
Thisworkwassu
pp
ortedb
y
theNationalNaturalScienceFoundationofChina
(
,
)
andtheChinaPostdoctoral
ScienceFoundation
(
M
)
 
:
(
wan
g
xinhuasdnueducn
)
 
 
 
 
(
 
 
)
(
 
 
)
(
lei
g
uocs
g
mailcom
)
SharedAccountCrossDomainSe
q
uentialRecommendationwithSelfAttention
Network
GuoLei
,
LiQiu
j
u
,
LiuFan
g
􀆳ai
,
and Wan
g
Xinhua
(
Schoolo
f
Business
,
Shandon
g
NormalUniversit
y
,
Jinan
)
(
Schoolo
f
In
f
ormationScienceandEn
g
ineerin
g
,
Shandon
g
NormalUniversit
y
,
Jinan
)
Abstract Sharedaccountcrossdomainse
q
uentialrecommendation
(
SCSR
)
isthetaskofrecommendin
g
nextitemsina
p
articularcontext
,
whereusersshareasin
g
leaccount
,
andtheirbehaviorrecordsare
availableinmulti
p
ledomainsCom
p
aredwithtraditionalse
q
uentialrecommendationtasks
,
SCSRis
challen
g
in
g
dueto
:
)
Theinteractions
g
eneratedb
y
anaccountisamixtureofmulti
p
leusers
)
The
behaviorsinonedomainmi
g
htbehel
p
fultoim
p
roverecommendationsinanotherdomainRecentl
y
,
mostoftherelatedworkisbasedonrecurrentneuralnetwork
(
RNN
)
Duetotheinherentdrawbacks
ofRNN
,
RNNbased methodsaretimeconsumin
g
and moreim
p
ortantl
y
the
y
failtoca
p
turelon
g
ran
g
ede
p
endenciesofaccounts􀆳interactionsInthiswork
,
wetar
g
etatSCSRand
p
ro
p
oseaself
attentionbasedcrossdomain recommendation model
(
SCRM
)
to address these two challen
g
es
S
p
ecificall
y
,
to modelthe mixedinteractionsfrom multi
p
leusersofasin
g
leaccount
,
a multihead
selfattentionnetworkisfirstintroducedThen
,
tolevera
g
ethedomaininformationinonedomainto
im
p
rovetherecommendationinanotherdomain
,
thecrossdomaintransfernetworkbasedonamulti
la
y
ercrossma
p p
erce
p
tualnetworkisinnovativel
y p
ro
p
osedFinall
y
,
a h
y
bridrecommendation
decoderisex
p
loredtoconsidertheinformationfrombothdomainstoachieverecommendationineach
domainWeconductex
p
erimentsonarealworlddatasetHVIDEO
,
andtheex
p
erimentalresultsshow
thatSCRMout
p
erformsallthebaselinemethodsintermsof
MRR
and
Recall
Intermsoftrainin
g
efficienc
y
,
SCRMachievesshortertrainin
g
andlearnin
g
timethanRNNbasedmethods
Ke
y
words  multiheadselfattention
;
sharedaccountrecommendation
;
crossdomainrecommendation
;
se
q
uencemodelin
g
;
collaborativefilterin
g
 
 
(
sharedaccountcrossdomainse
q
uentialrecommendation
,
SCSR
)
使
,
,
:
)
,
;
)
,
(
recurrentneuralnetwork
,
RNN
)
,
RNN
,
RNN
,
SCSR
,
(
selfattentionbasedcrossdomainrecommendationmodel
,
SCRM
)
,
,
,
,
HVIDEO
,
,
,
MRR
Recall
;
,
RNN
 
;
;
;
;
 TP
;
TP
  
,
,
,
,
,
[
]
广
,
广
,
(
sharedaccountcrossdomainse
q
uentialre
commendation
,
SCSR
)
,
:
使
,
使
(
);
[
]
,
:
)
使
,
使
,
使
,
,
,
使
)
:
,
,
,
,
(
)
,
使
,
,
,
使
(
RNN
)
(
convolutionalneural
network
,
CNN
)
[
]
,
RNN
CNN
,
,
RNN
CNN
,
,
S
{
,
,
,
T
,
,
},
,
,
,
,
,
,
π
Net
[
]
,
,
:
)
π
Net
使
RNN
 
:
of 14
免费下载
【版权声明】本文为墨天轮用户原创内容,转载时必须标注文档的来源(墨天轮),文档链接,文档作者等基本信息,否则作者和墨天轮有权追究责任。如果您发现墨天轮中有涉嫌抄袭或者侵权的内容,欢迎发送邮件至:contact@modb.pro进行举报,并提供相关证据,一经查实,墨天轮将立刻删除相关内容。

评论

关注
最新上传
暂无内容,敬请期待...
下载排行榜
Top250 周榜 月榜