Tải bản đầy đủ (.doc) (73 trang)

Đồ án tốt nghiệp Nghiên cứu các kỹ thuật phân tích câu hỏi và sinh câu trả lời

Bạn đang xem bản rút gọn của tài liệu. Xem và tải ngay bản đầy đủ của tài liệu tại đây (1.07 MB, 73 trang )

MỤC LỤC


1. Giới thiệu chung về hệ thống hỏi đáp 7



 !"
Báo cáo đồ án tốt nghiệp
DANH MỤC CÁC TỪ VIẾT TẮT
STT Từ Dạng đầy đủ Giải nghĩa
# $ $%&'()*'+&,) '(&/ 0(12)345
 6)- ''%&768'&9:*,/8()* '(&/ 0(12)345;<=3>
 ? 8(%,8@8%8&?,*<&'')
A(%B(CD@EFG(H
)I
J K KC5&<(&9'+&,.5& L<M%(,N@O)/*3P)
Q RK R&%@8,KC5,&'')* 6)S%(T<<UV%.
W K )8@*%&*X&K)&
 Y$ Y7$%&'()* M%2)<Z([3S2)
\ ]$ ]&'*$%&'()* M%2)3^_'8)
` $ @(&,8()X&$%&'()* M%2)@H8<a
#b ?' )8@*%&?8,()<)58(' O)(8/)8c)(*L)
## - 'd9G@)0%
# - *(8@:*,/8()*-(8(& e(<N<4<(,L(4)(F()
# - :*,/8()*-(8(& ,L(4)(F()
#J $ $%&'()*'9&,)'<%'')* M%2)(N*@%B
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 2
Báo cáo đồ án tốt nghiệp
DANH MỤC CÁC HÌNH VẼ
f#)g(,^</c(0(1$\


fP<3h0(12)345##
f0(1$#W
fJi8<=8([j')5jb
fQe%(,^<(,)&C4<3k(H<(S
fW%.0((,)&J
fe%(,^<(,)&lFC4<3k@*L)(H<(SQ
f\)g(,^<6)-##
f`4<(m5n<=8-J
f#bR&'*%,<&)(&,:8<&'_<4<o)g(m)%.IJQ
f##-3hl1)<*X)0<5M(U<<M%2)'D9p(,)&W
f#m)3q((%B((*49%.0((,)&WJ
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 3
Báo cáo đồ án tốt nghiệp
DANH MỤC CÁC BẢNG
6N#-Hl4<8%)G8$Xm0(1(f/l)g/
6N4<<M%2)/r%
6NF()X><4<*9&(,*(,)&
6NJ1)V%80,&'*@X&')G8<M%2)_<M%(,N@O)\
6NQ1)V%80,&@&X8()G8<M%2)_<M%(,N@O)`
6NW03>(%3P<l)lg(P5<M%2)7<M%(,N@O)Jb
6Nc('1<M%2)XU9pl)<m)3q(<(,f9s(,)&QW
6N\^(U<<4<KQW
6N`^(U<*L)@0QW
6N#b6NR-777%(,G(F()*L'iQ
6N##6N?_F()X>(,8(,*5t(,)S@u/Q`
6N#6NvKRK_F()X>(,O54)c)*LQ`
6N#6N"YKR777%(,G(F()<=8O)<=<Z(,83P<(,)S@u/Wb
6N#J4<,mo%c<(,*@iXH<5t(,87X)0oN*(mW\
6N#Q4<([3hi8(,*@iXH<5t(,87X)0oN*(mW\
6N#W1)@)I0)G8<4<<M%(,N@O)wXm<4<9)8@*%&/*X&(,*@iXH<5t

(,87X)0oN*(mW\
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 4
Báo cáo đồ án tốt nghiệp
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 5
LỜI CẢM ƠN
,;<g(&/C)<M(mD)@O)<N/'M%'w<(;)<4<(n.<F
(,*(,OxL)a<64<l*8mc)Z)<%Xm<4<(n.<F(,*l*8
F0(F()yoc/F0(1(F()Z),)I3u(B(f)N
9L.y(,%.>3L(<*&/Gl)g(T<yGl))0/V%Eo4%(,*
'%1(V%4(,fz/z/a<(B5Xm,{@%.0(L)(,O
K/C)D)@O)<N/3g<FI8_)NX)Ioc/F
0(1(F()yl*8F0(F()y(,OxL)a<64<l*8m
c)3ug(@t)^53|y;9rXm<}9L.(B(f(,*V%4(,f&/
(H<)0xh4(1()05
%1)<s&/C)<M(mD)@O)<N/3g)83fyoLo{3u
3cX)Iy<z/'Z<y3ZZ5El)gXm)^53|(,*V%4(,fa<(B5y
)I<T%Xm*m(mxh4
Báo cáo đồ án tốt nghiệp
TÓM TẮT NỘI DUNG ĐỒ ÁN TỐT NGHIỆP
sX;)'H54((,)SlF[<=8l)(g7Cuc)y<~l*8a<lA(%B(y
@P(F()m.<ml•@hy(g);)o;<Xm*l€%.I/;)yl€%.IX>'Hos
•(F()'1•3Z8)w/ow(3P<)>%(F()(f<c)3S(,d(m‚O)<)g
(wƒ'„<8*UXfXB./m%<n%(f/(F()8ay<UC4<Xm()0@P)
m./c(<8*
xS345T%<n%3Z(8<nCM.9H/c(0(1(,U<,^((F()@E(d@m0
(1<ZlNz(H<)0<4<)8*()g5y)S%FG<*O)Z<*5…5O)
'D9p38,8.I%<n%(F()yG<M%2)9;)9LFG(H)I†8(%,8@
@8%&‡y0(15N))S%3P<<4<(F()3Zd/T<3c<*O)ˆl)B3P<
.I%<n%(F()8.<M%2)yZ5N)(,NX>3^(F()O)'D9p<nylF()g%y
lF([8

)08.y/q<9s<8<Z/c(0(1(,U<,^((F()m*3L(3g/T<3c@E(d
<*O)/*/%13u<Z/c('10(154((,)S3L(3P<Glg(V%N
lNV%8xZ<U@m0(12)345†$%&'()*'+&,) '(&/7$‡
,*3h4(1()05m.o8*h/<4<Xe3>'8%ˆ
− f/)S%X>l)g(,^<<=80(12)345
− ?M(U<<M%2)9H8Xm*(,)&
− f/)S%X>/c(0(1<p(S@m6)-#†)''%&7o8'&9):*,/8()*'.'(&/70
(1c)(*L);<=3>‡
− m)3q(
+ ?M(U<<M%2)9H8Xm*(,)&T9p(,*oN*(m
+ m)3q(0(1c)(*L)T9p(,*oN*(m
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 6
Báo cáo đồ án tốt nghiệp
Chương I. TỔNG QUAN
1. Giới thiệu chung về hệ thống hỏi đáp
,*@iXH<(,U<,^((F()y0(12)345†$%&'()*7'+&,)yX)g((w(
@m$‡3N/BX)0<(H3c(,N@O)<4<<M%2)3P<B5Xm*d9LFG(H
)I0(19H8Xm*<'d9G@)0%8.<4<XzoNFG(H)I<Z'‰y38
,8<M%(,N@O)5sP5(&*FG(H)I
$3t)2)<4<lA(%B(CD@EFG(H)I†8(%,8@@8%8&5,*<&'')
7?‡<4<9L(,U<,^((F()l4<yXU9pˆ(,U<,^(XzoN†9*<%/&(
,&(,)&X8@‡
$@m>/Z3S54((,)S0(1c)(*L)ylg(P5$X;)<4<<<gV%N
@Ec)(*L)†)8@*%&88&/&(‡y(8'„3P</c(0(1c)(*L)*m<}y0
(1XmO)9s3>%)8*()g52)_(,N@O)
4<(,U<,^((F()(%.>(1@m9H8Xm*E(d<*)(B5XzoN
GZ/([,)Io)0(yO3Zyc)9%Gi8<=8XzoN(3X;)
(B5([l*4†l&.+*,9‡Z<T8/mlFV%8(M/3g(,B((H([8./1)V%80
(,*<M%%.)Iy<}<Z([l*4(F)(f<83=y(,B((H([XmV%80G545,e(
V%8(,a(,*X)0<V%.g(3ki8<=8<M%0(12)3459H8(,I<4<()g5

<B<^(,a3gG545<M%(FV%8X)0<<*5…5O)9sB5<M%2)
9;)9LFG(H)Iy38,8<M%(,N@O)wa(&*FG(H)I
Bảng 1. Sự khác nhau giữa QA và hệ thống tìm kiếm
$ -KR
5%( M%2)FG(H)I %Š)(,%.Xe<T8([l*4
"%5%( M%(,N@O)wa c((B5P5(m)@)0%
'&' ,8<T%8 )I<T%/d,c
H<(g3u<Z)>%0(12)3453P<CM.9HyX><e%(,^<<=8<4<0
(1m.<Z(Sl4<8%yf<%y3>%<ZGXe3><U<nV%8
(M/CD@E@mˆ
• ?M@*L)<M%2)
• ‹D@E<M%2)
• •<T<<'d9G@)0%
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 7
Báo cáo đồ án tốt nghiệp
• ,U<,^((F()([XzoN
• -)<M%(,N@O)
Hình 1. Kiến trúc một hệ thống QA
c(Xm)0(12)345(H3c()I%o)S%(,*(H<(gˆ
• ?)$-'*†((8,9)&(8@ybb#‡
• 'l-R†6,)@@&(8@ybb‡
• RK†&C(RK(,)&X8@*:&,&<&y([#``b3g8. )
?)$-'*@m0(12)3459H8(,I'Hlg(P5<=8<4<lA(%B((,U<,^((F
())03L)X;)<4<oc@a<FG3S@H8<a3*LXzoN<T8(F()5sP5
3S(,N@O)6c@a<FG3P<9H8(,I,e()>%<F<pCD@EFG(H)I
ˆ?"-(8&,†Part-Of-Speech Tagger7B9)0@*L)([‡yK(8&,†Name Entity
Tagger_B9)0(H<(S‡Xm<N-X>FGZ5M(U<<M%2)3S(f/,8
([lZ83S(,%.XeXm(f/,8Ky5M(U<<4<3*LXzo8*h/<NX)0<l)S/(,8'H
<Z/q(<=8<4<(H<(S<=8KXm/1)V%80)G8<4<([
'l-R@m0(12)345<Z3c<UC4<<8*9H8Xm*X)0<5M(U<@P

9G@)0%*@)&@;/mZ(%3P<@m9H8Xm*<4<lA(%B(CD@EFG(H)I
xM.@m0(13)N)0%V%N
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 8
Báo cáo đồ án tốt nghiệp
RK(H<)0@%(,G<4<<M%2)3P<5M(&*([@*L)<p(S†XU9pˆ(•
<T<y'Hl)0yŒ‡)0<5M@*L)<M%2)Xm<M%(,N@O)9H8(,I(f/(F()([/c('1
@P@;<4<XzoNy(m)@)0%
2. Khó khăn trong xử lý ngôn ngữ tự nhiên
2.1. Cấu trúc ngữ pháp và các thành phần cấu tạo câu
,*)I<T%X>G545()g)0(<^(8<Z(S(e.,•<8<Z/c(
3ki8<%ŽX><4<a)<=8@*L)([<~G545<M%()g)0(
,*()g)0((,B((H'w53q(<4<([@m5()0<U3So)S%(kV%80
G545_(T<@mV%80<^545_)G8<4<([(,*/c(<M%,B((H<4<.g%(1<e%
(m3P<V%.3ko•/c(Xk(,Ue(3k)Xk(,U(8.3•)(fi8<Z(S(8.
3•)(&*
[(,*()g)0(<Z(S<Z/c(()g8.h/)>%()gG8y(,*
()g)0(yG([)>%()g@L)<Z(S3P<…5od)G()g8.([l4<<Z
i8U9p8)([/c(()g3e(y;<<Z(S3P<…5X;)8%(m/c(([<ZE
i8(,[%(P@m3e(;<
*m),8<t<Z<4<<p/98([y3c([Xm(U([X;))>%(m5n<e%(L*y
3N/)0/<4<<T<zl4<8%(,*<M%
2.2. Phép lặp từ
4<3L)([y3L)(['dG%8.98([(O3P<'D9p3S<}3gG(,)
(T<3u3P<w<3g(,*<4<<M%2)(,;<y5…5G545XB.a)@m5…5@q5
([†885*,8‡‹…(3*Lc)(*8k'8%ˆ
> Quyển sách bìa đỏ kia là sách tham khảo à?
Đúng vậy.
> Nó bao nhiêu tiền?
[Quyển sách màu đỏ kia] bao nhiêu tiền ?
20.000 đồng.

0(15N))S%3P<‚Zƒd3M.<U@m‚V%.S'4</m%32l)8ƒc(
55453)N3SCD@E@*L)<M%m.@m@%(,G98'4<<4<(,)(T<3u3P<
I%(,;<3Z)q55N)/c(3L)([y0(1l)S/(,8@L)98'4<([3)S//;)
e(Xm(8.(g3L)([X;)(,)(T<ne((28/uG545Xm);)L<M%?…5
@q5([@mXe3>5•o)g(,*CD@EFG(H)I
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 9
Báo cáo đồ án tốt nghiệp
2.3. Câu tỉnh lược
mm.y<*O)(OC%.I'D9p<M%lF*m<}†<M%(}
@P<‡(,*c)(*L)y/mi8<=8G<M%3Z5p(%c<Xm*G3*L2)345
(H<)0(,;<3Z
‹D@EG(,OP55T<(L5<=8<M%(}@P<3t)2)0(15N)9%.(,f
/c(/*9&@c)(*L)y3)>%m.lF5N)@^<m*<~9•9m/q<9sX)0<ow(o%c<
5N)2)<M%2)3n.3=l4@mlZ<k%X;)O)'D9p
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 10
Báo cáo đồ án tốt nghiệp
Chương II. KIẾN TRÚC CHUNG HỆ THỐNG HỎI ĐÁP
3uZ)d5n(,;<y<4<0(12)345(H<(g<Z(S<Z<e%(,^<'8)l4<
8%3>%5N)(B5(,%CD@E<4<Xe3><oN
• ?M@*L)XmCD@E<M%2)
• •<T<<e%(,^<9G@)0%
• 'd9G@)0%
• ,U<,^((F()
• ‹D@EXm')<M%(,N@O)
Hình 2. Lược đồ hệ thống hỏi đáp
,*<m.y(8'„(B5(,%(f/)S%l)g(,^<<=8/c(0(12)345Xm
3g<(8'„(f/)S%(I/X><<gV%N@Ec)(*L)y([3Z<m)3q((D
)0/0(1c)(*L)
>l)g(,^<<=80(12)345y<Z(S<)8(mZ/<U3S()g5<Bˆ
Z/@)IV%83gCD@E<M%2)Xm<M%(,N@O)†):*,/8()*&C(,8<()*•'.(8<()<

88@.')'‡yZ/@)IV%83g(,%.B5y(,U<,^((F()(&*([lZ8†):*,/8()*
,&(,)&X8@‡f(gy(8'„()g5<B0(12)345(&*0(1<*(,I
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 11
Báo cáo đồ án tốt nghiệp
1. Kỹ thuật nhận dạng câu hỏi và trả lời
)/c(<M%2)3P<3q(,8<*/c(O)8./c(0(1(,N@O)(H3cy
)0/Xp3n%()I5N))N)V%.g(•/38,83P<<M%(,N@O)@m5N))S%3P<<M%
2)
$%4(,fCD@E/c(<M%2)o8*h/<4<o;<<*'8%ˆ
 ?M(U<<M%2)
 ?M(U<G545<=8<M%yX)g(9L<M%(,N@O)
 a<9L<M%(,N@O)
 a<<4<V%805sP5
 -w5Cg5i8(&*C4<'%e(
 d,c(,%.Xe
$%4(,f5M(U<<M%2)o8*h/X)0<5M(U<<^545<=8<M%2)yC4<3k
@*L)<M%(,N@O)/*3P)y(f/([lZ8<Z@)IV%83S(H<)0(,U<,^(3*LXzM%
(,%.Xeo83n%CM.9H([G([lZ8XM.•//p<()I%3L(3c<UC4<
<8*Xm<ZlNz(,U<,^(,8/c('1GTX)I<*<M%(,N@O)(U<P5(,*
(B5P5@;†V%4(,f@a<‡
$%4(,f@a<9L<M%<Z)0/Xpl)S/(,8<4<TX)I<M%(,N@O)<Z<T8
G(H<(S<n<Zd<M%(,N@O)/*3P)8.lFXm@*L)o2G<M%lF
(U<P56c@a<<M%<Z)0/XpBo)g(<4</1)V%80)G8<4<(m5n([
lZ8<=8<M%2)yXm(f/l)g//1)V%80((H(,*GTX)I<M%(,N@O)
x)>%m.(O3P<(H<)03H8Xm*/c(<M.5p(%c</m<e%(,^<9*0(1
V%.g(3k‚*N<4<ƒ)G8<M%2)Xm<M%(,N@O)3P<(U(*43SC4<3k(,B(
(H<4<TX)I<M%(,N@O)M%(,N@O)3P<34)4<8*e(@m<M%(,N@O)3P<0
(1@H8<a
1.1. Phân tích câu hỏi
4<Xp5M(U<<M%2)<Z(SlF,‘,m31)X;)<*O)l)(,N@O)<M%

2)y(%.)I31)X;)/c(0(1(,N@O)(H3cZ3ZX8)(,t<1(.g%V%.g(3k
3c<UC4<<=80(1
w(@a<(F()Ž<T8(,*<M%2))GX8)(,tV%8(,ay<zoNy<*5…5
0(1(H<()3^@03S38,83P<<M%(,N@O)3^3w<*<M%2)3Z)
X)0<5M(U<<M%2)(e(oL)y,e(lZXmn@mlF(S<*0(12)345
(H<)0)0/Xp<=8Z
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 12
Báo cáo đồ án tốt nghiệp
4<()g5<B<zoNe(3S5M(U<<M%2)@m<)8)0/Xp(m5nˆ
(f/l)g/9L<M%(,N@O)/*3P)Xm(f/l)g/(,a(M/<=8<M%2)Re()>%<4<
0(1†*@@47@)*9bby&&(8@bb#Œ‡'D9p/c((B5<4<@%B(9*O)@B5
(,f(H(L*3S(f/l)g/9L<M%(,N@O)/*3P)†K‡O(f<4<@%B(3P<X)g(
9;)9L<4<o)S%(T<<UV%.†RK‡Xm)0/Xp(f/l)g/9L<M%(,N@O)@mX)0<
l)S/(,8'Ht8P5)G8<M%2)Xm<4<o)S%(T<3Za)o)S%(T<3>%((U<
X;)/c(9L<M%(,N@O)/*3P)Xm3P<w<*<M%2)5sP5X;)3k9L
4<()g5<B3)Ne(3S(f/3P<(,a(M/<=8<M%2)3Z@m@*L)o2(e(<N
<4<([9[†lF<Zi8V%8(,ay<}<Z(4<9pX>/q(G545‡l2)<M%Xm
<*)G([<t@L)3L)9)0<*(,a(M/<M%2)†([lZ8‡
c((,*GlA(%B(3P<'D9py3Z@m'D9p<M.(,)&†/c(9L<M.5p
(%c<‡e3>m.'„3P<@m/,‘(,*<'8%
Trích rút từ khóa (keyword)
xM.@mo;<3n%<*X)0<@H8<a([G3S')<M%(,N@O)5sP5
0(1<*)([lZ8@m(e(<N<4<98([y3c([y(U([y5Z([(,*<M%2)y
@*L)(,[98'4<<4<([o8*h/†V%.g(3k9H8Xm*l))0/<=8O)()g(
lg‡4<([/m/*9%@&5M(U<<^545lF<*)@m/c(5n<=8([y<p/([F
Ga<<~ok@*L)o2
)0<5M(U<<M%2)(,*0(1$3ZX8)(,tV%8(,a*9%@&5M
(U<<M%2)†V%&'()*88@.')'‡<*(83n%,8@m<4<(F()ˆ
• xk9L<M%(,N@O)/*3P)
• *L)<M%2)

• B5([lZ8
1.2. Luật phân tích từ ngữ
4<()g5<B0<=80(1$9H8(,IG<F<pCD@EFG(H
)Iy/m3c<UC4<Xm<e(@Pok);)LXm5p(%c<Xm*l)g(,^<(•(S
<=8(*m0(1xZ@mG<F<p))58,yY*,9&(y8/&K()(.(8,&(Œ
,*FG(h(L)G([<Z)>%i8*q<)>%([3hi8yn
i8ŒfXB.y(,*0(1<nCM.9H(,;<G@%B(3SV%.g(3k([@*L)
<=8([3S5p<Xp<*V%4(,f5M(U<<M%2)Xm')<M%(,N@O)c((,*G
<4<()g5<By3Z@mCM.9HY*,9&(
1.2.1. Xác suất ngữ nghĩa (Sense probability).
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 13
Báo cáo đồ án tốt nghiệp
,*V%4(,f5M(U<<M%y0(1(O5N)V%.g(3kC&//c(([<Z
(%c</c(@*L)([e(3km*lF†(,*V%4(,f(f/9L<M%(,N@O)/*
3P)‡x)>%m.<Z(S(H<)03P<9H8Xm*(U(*4C4<'%e(Gi8<=8([•/
(,*/c(9LY*,9&(†XU9pC4<'%e(<=8([‚<8(ƒ•/<8*d/p<‚8)/8@ƒ
(,*Y*,9&(‡Y*,9&('w5Cg5i8<=8([(&*<%Š)U9py<*/c(98'4<
'w5Cg5Gi8<=8([+@m’'byŒy'“(8(U(*4C4<C%e(i8<=8([(%c<Xm*
/p<'8%ˆ
;)
ml@m/c(o)g5M<e5no•C4<'%e(<=8i8Y*,9&(3^3n%()I
†l*Nb‡
1.2.2. Dạng của từ
L<=8([+3P<(Uo•ˆ

1.2.3. Khoảng cách của từ (Word Distance)
)0<(U(*4l*N<4<<=8([3P<'D9p3S(U(*4l*N<4<)G8
<M%y<p(Sy)G83*L(,N@O)Xm<M%2)
*N<4<)G88)([(,4)i85p(%c<Xm*l*N<4<Gi8d<s
3c'M%(,*5M@*L)Y*,9&(I@0<3c'M%<=8(e(<N<4<<q5Gi8<=8

8)([3P<3*3L<9H8Xm*C4<'%e(<=8<q5Gi8<=8([(U,)IXmlg(P5
@L)
*N<4<)G88)([3hi8<~3P<(U(*49H8(,I(e(<N<4<G
i8<=8([y(s.(&*C4<'%e(*N<4<)G88)([ylE)0%@m9)'(†+#y+‡3P<
3ki8@ml*N<4<3hi8*q<(,4)i8(s.Xm*([3Z@m3hi8
8.(,4)i8
1.2.4. Thay thế từ ngữ
)0<(8.(g/c(([m*3Z<Z(S<n()g((,*V%4(,f/d,cV%&,.y3P<
(U(*49H8(,I(B5Y(e(<N<4<(B53hi8<=8([+B5<4<([(8.(g<*+
3P<3ki8@mˆ
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 14
Báo cáo đồ án tốt nghiệp

,*3Z‚(ƒ@m/c();)L<13ky'D9p3S(,4(,OP5/c(([n
i8X;)+<ZGi8lF(H<'HwX;)+†XU9p‚<*/5%(&,ƒX;)
‚/8<)&ƒ‡
1.3. Luật viết dạng câu trả lời - query reformulation & query expansion
1.3.1. Viết lại dạng truy vấn - Query reformulation
)/c(<M%2)3P<3q(,8y0(1'„(H3c'),8/c(*q<Xm)<%Š)<Z
lNz@m<M%(,N@O)<*<M%2)
x1)X;)<4<<M%2)(,*()g)0((f<4<(1(e(@m)G%.I<e%(,^<<M%y
(8.Xm*Xk(,U<=8([3S2)o•@O))N)
U9pˆ‚4<)N<=8oT<(,8)g%GoI*8%0@m8)”ƒ
M%(,N@O)3)Ne('„@m‚4<)N<=8oT<(,8)g%GoI*8%0@m
•K–ƒ
4<(T8)@mX)g(@L)<M%ˆ‚•K–3uX„oT<(,8)g%GoI*8%0ƒ
xS()gm(,N@O)3P<(&*<4<(T(85N)(I/Xm*3c(['8*<*5s
P5X;)c)9%<M%2)/q<9s<M%2)lF<T83c([m*UXfXB.I
<M%(,N@O)(1(e(<*<M%2)(,*()g)0(@m(&*9L#
1.3.2. Mở rộng truy vấn - Query expansion

4<([l*4(f/3P<(,*V%4(,f5M(U<<M%2)3P<38Xm*V%&,.$%.
(,fm.<n()g(3S<Z(S(H3c(,U<,^(9G@)0%')<M%(,N@O)K
6;<3n%()I(,*X)0</d,cV%&,.@mX)0<5N)CD@Ef(4)38Gi8
<=8([Gyo•X)0<(8.(g<4<([l*4o•G([3hi8(T†(,*
(05<'d9G@)0%‡g%<n()g(  C4<3k(,a(M/<M%(,N@O)c(Xm)()>(1
(OC%.IC%e()0(,*<M%2)‚Xk(,Uƒy‚dƒy‚(L)ƒ/*3P)<M%(,N@O)X>
3k83)S/—‚m.ƒy‚@^<ƒy‚l)ƒ/*3P)<M%(,N@O)X>(O))8Œ-8%CD@Ey<^ok
@*L)o2l2)V%&,.
t()g5(&*y(8<n/d,cX)0<(f/l)g/o•X)0<(I/Xm*†(,*(,O
P5‚*,ƒ‡<4<([3hi8X;)<p/([<n(f/l)g/
6;<()g5(&*@mX)0<CD@E<4<(,L([yo•G/c(<4<(U<P5Xm@*L)o2
<^l2)G<p/([<n(f/l)g/†CD@E‡()g5
,*XtCD@E<%1)yg%V%&,.<t<T8)>%([l*49;)9L@)I
([y3c([<~(IO)†g%<Z<NaXm(I‡ok@*L)o2g%8.<N'8%'H<w(
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 15
Báo cáo đồ án tốt nghiệp
(}8XB./mXr<t(h(L))>%([l*4@)I([y<^(8<~'„@*L)o21(
G([l*4@)IV%8(,H<()g53gG([l*43u3P<CD@Ed5n(,;<†<Z
(S(FV%8<M.5p(%c<3S5M(U<<e%(,^<FG‡
2. Kỹ thuật trích rút dữ liệu - retrieve document & answer extraction
2.1. Trích rút thông tin - Retrieve document
i)Iy3)>%lF(S()g%(,*/c(0(12)345(H3c@m5N)3k
Xk3P<(B5(m)@)0%<T89G@)0%/m<M%2)<n<Z,*V%4(,fm.y'H'8)l4<
)G8<4<3k9L<M%2)Xm<4<TX)I<M%(,N@O)<n3P<C&/C…(y'H@H8<a
5N)3=@;3S3N/oN*3c<UC4<<8*5N)l)S/'*4(3P<'H)•%9*X)0<
X)g(@L)3k9L<M%†,&:*,/%@8()*‡0(1$3u(4)'D9p0(1(,U<,^(
(F()3S(H<)03)>%m.
U9p 0(1$†$%&'()*7'+&,)5,*,8/*:(&8%8&89
*)()*‡ˆ
Hình 3. Hệ thống QALC

Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 16
Báo cáo đồ án tốt nghiệp
*9%@&5M(U<<M%2)V%.g(3k<4<(F()ˆ3k9L<M%(,N@O)
/*3P)yZ/<=8<M%2)y([l*4ŒF()m.'„3P<'D9p3S(,U<,^(9G
@)0%(FV%8oc(f/l)g/
F()8.3*LXz/m<M%(,N@O)38(f/l)g/@mlg(V%N<=8/c(Xm)V%4
(,f<a@H8(m<Fyh/<Zˆ
• H8<a3n%()I9H8Xm*<4<([<ZEi8†([lZ8‡<=8<M%2)(f/l)g/(05
9G@)0%
• )gm()g5'H@H8<a(T(FV%8X)0<B9L<4<<p/Gi8
3hi88.38i8Œ
• 4<<M%3P<@H8<aO;<@P3c<I@0<‚l*N<4<ƒGi8
)G8<^X;)<p/([<=8<M%2)Xm(I(H<(S(,*9L(,N@O)/*3P)
xn%,8<=8/*9%@&m.(OlF5N)@m(U<P5e(<*0(12)
345ˆ(m)@)0%<Z(SlF3P<Cg5Lg%@m<M%2)3^7'8)y'H@H8<a9H8
)>%Xm*([l*4@moN(M<M%2)Œf(gy(O3n%,8'„3P<CD@E()g5d
o;<'8%@m(,U<,^(<M%(,N@O)†8'+&,&C(,8<()*‡
2.2. Trích rút câu trả lời - Answer extraction
,U<,^(<M%(,N@O) - '+&,&C(,8<()*y8.<ta)@mo;<CD@E<^545y@m
o;<()g5(&*'8%X)0<(,U<,^(9G@)0%3S38,83P<<M%(,N@O)/*/%1L)
o;<m.y<4<<M%(,N@O)()>/z3P<(4<,8Xm34)4
8)<4<5•o)g3S(,U<,8<M%(,N@O)y3Z@mˆ
− ?M(U<9L<M%(,N@O)†'+&,(.5&88@.')'‡
− )g5<B(&*;a</4.†?8((&,@&8,)855,*8<‡
Phân tích dạng câu trả lời -Answer type analysis:
 ‹M.9H(B5<4<9L<M%(,N@O)
 ?M@*L)<M%(,N@O)
 4<l%F9L3P<wXZ)9L<M%(,N@O)(T
 4<m.<Z(U<UC4<<8*lF(,N@O)3P<<4<<M%2)*m)
l%F9L†lF4u3P<‡

Tiếp cận theo hướng học máy -Pattern learning approach
 L<M%2)<Z(S3P<‚a<ƒg%'D9p<q5<M%2)7(,N@O)/c(l)S%
9G@)0%@)I(p<
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 17
Báo cáo đồ án tốt nghiệp
 %B((*4
− %F<T831)(PyXz<NXm<M%2)3P<38,8oc(f/l)g/U9pˆ
<8@@*,)&'y6)8<yQWb
− e(<N<4<(I(H<(S3P<(8.(go•9L<=8<^
− c(Xm)l%F9L<s@*L)3P<9s3S‚a<ƒX>Xm)9L'dG%3q<
(,
− H<)0(•V%4(*4†(&*9L(H<(SXm<4<@H8<a(*l&l4<‡
 -D9p‚3c()(dƒ3S34)4<M%(,N@O)xc()(d3P<(Uoz
<F(T<†3^˜3^™'8)‡
v)8)3*Lm.<Z(S3P<5M,u)8)3*Lm.(mJ)8)3*L<*
• f/,8<4<(H<(S7K()(.KC(,8<()*ˆ)0/Xp<=8)8)3*L@m,^(,8/c(
(B5TX)I<M%(,N@O)([(m)@)0%Z(S(e.8.,•yg%'D9p<M.(,)&y
V%4(,f9%.0(<M.lg((^<y(8'„(%3P<9L<M%(,N@O)/*/%1[
9L<M%/*/%1y'),8(B5P5<4<(H<(S'w5Cg5(&*<4<@%B(X)g(@L)
<M%†,&:*,/%@8()*‡
• ?M@*L)<4<(H<(S7K()(.<@8''):)<8()*ˆB5(H<(S(%3P<d)8)3*L
(,;<<Z(S3P<5MXm*/c('1Z/e(3k4<@*L)(H<(S<Z(S<Zˆ
+ O)yXk(,Uy(•<T<Œ
+ m.(4y(O))8
+ xk@Py'1@P
+ %lšy5…53*
Š)9L(H<(S.I%<n%<4<'w5<M%XmCD@E(T
• ?M@*L)(,%.Xe7$%&,.@8''):)<8()*ˆs3SC4<3k@%B(<^545<n
9%3SX)g(<M%(,N@O)([<M%2)3u3P<5M@*L)d)8)3*L(,;<
• -*'4<4<(H<(S7K()(.R8l)ˆ$%.g(3k9L(H<(S<Z5sP5

9L<M%2)lF6;<m.<Z(SCD@E<%l)5M(U<<M%2)g%'D
9p<e%(,^<<M.5M(U<-D9poc@a<<M%3SC4<3ky34)4<M%(,N@O)
((U<X;)<M%2)y<F(T<<*3)S/3P<@e.([RK
2.3. Kỹ thuật xử lý đại từ thay thế
U9pˆ)<M%2)'8%3P<B5Xm*ˆ‚6T<(,8(*e(3P<X„z/m*”ƒ
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 18
Báo cáo đồ án tốt nghiệp
0(1<n5N))S%‚6T<(,8(*e(ƒ@m3L)([(8.(g0(15N)()g
m(f/l)g/(,*-3S(f/,8(IoT<(,8<Z(%c<(UlU<(U<3L(@;
e(y'8%3Z(SXm*<M%2)y38<M%2)X>9L3)Nˆ‚6T<(,8›?3P<
X„z/m*”ƒ
2.4. Tổ chức cơ sở dữ liệu độc lập
'd9G@)0%<~@m/c(Xe3><nV%8(M/<=80(1,*5M(U<<M%
2)y3L)38'1<4<0(12)345)08.9H8Xm*3n%Xm*G545@m(B5<'d9G
@)0%<=8Y*,9&(XmRK%h9G@)0%@%(,G(F()y(%š0(1<Z(S@m
<8'w5Cg5*q<'w5Cg5/c(5n4<0(1<'d9G@)0%<%.I)05)0
8.-$'&,X&,y",<8@&'D9p@%(,G<'d9G@)0%(,I<4<oN
e3>3q(,8@m<4<(T)G8GoN9G@)0%m.X;)0(1<^
545Xmoc@a<<M%5sP50FG(H)I3uZ)d5n(,IZ(S(8/lN*
<4<(•<T<<=8Kœ†K97'&,œ,)&9@.(&,:8<&(*8(888&/&(‡y(L)3Z(•
<T<9G@)0%3P<(•<T<3c<@B5X;)5nT9pG@)0%(,*G0(1
@)IV%83gFG(H)I(Oo8*h/(m5n@mˆ
• 4<oNGi8G(H)I
• 4<oN@%(,G9G@)0%
• CL)G8oNG545XmoN9G@)0%
2.4.1. Bảng ngữ nghĩa câu - Semantic tables
xM.@m@*L)9G@)0%(S)0<4<f<=8<%.I)8X>/q(FG(H)Iy
(L)3Zy(B5P5([3P<54((,)SXm<e%(,^<<M%<oN3P<3ki8[G3P<
5M@*L)ˆ
• 4<(H<(S

• -Hl)0
• T<z
• c(5n*q<(*moc<M%(mGy(p<G
• m5n@)Ilg(
• $%4([
• xL)([
• G([38<^<z<^545
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 19
Báo cáo đồ án tốt nghiệp
c((H<(Sn((H/c(98([Xm'Hl)0(f(OwX;)/c(
3c([m5n@)Ilg(@m()>(1'„3P<@*L)o2'8%l)<M%3P<5M(U<
$%4([†'.'(&+*,9‡y@)I([yo•i8Z55nV%.g(3ki8<=8<M%
lF@)IV%83g5n9G@)0%<zoNxL)([@mG([w<@L)<*<p/([3u
3P<w<3g(,;<3Zy3L)([@mG(['„ok(8.(g(,*V%8(,f5M(U<<M%
Z/@L)y<}<ZG([@)IV%83g9G@)0%<oN@m([(%c<Z/(H<(Sy'H
l)0y<T<z
H<(SXm'Hl)03P<3ki89H8(&*<e%(,^<‚<8'&ƒ†(f%1@H8
<a‡o)S%(k'Hlg(P5)G8<^545XmGi8?n(D([XH/m<Z(Slg(
P5X;)/c((H<(Sl4<(L*(m<p/98([y*q<lg(P5X;)/c(3c([(L*
(m<p/3c([y'„(%c<(,OP5(H<(S*q<'Hl)0c(‚<8'&ƒ<Z(S3P<
349e%@mow(o%c<*q<(%š<aGi8<=8/c((H<(S*q<'Hl)03P<
3ki8od)/c((B5<4<(,OP5†‚<8'&ƒ‡y58o)0(9L3c([Xm9L98
([U9pyo8i8<=8([‚')5ƒ3P</)*L9;)3M.ˆ
Hình 4. Nghĩa của từ "ship"

Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 20
Báo cáo đồ án tốt nghiệp
2.4.2. Bảng dữ liệu cơ bản
6N9G@)0%<oN(S)0<e%(,^<9G@)0%(&*<4<f<=8/c(0(1
V%N@E9G@)0%f(e(<N<4<0V%N(,k<'d9G@)0%CD@E<4<(m5n9G@)0%

o•<4<(•<T<G‚Z/ƒ@)Ilg((H<(S)0<<Z/c(3k9LoN<%
@m*m(*mlN9iX;)X;)oe(lš0V%N(,k<'d9G@)0%m*
6N9G@)0%<oNh/(m5ny/c(@*L)<T8<4<(F()<=8‚Z/ƒ
Xm<4<(m5n9G@)0%Z/3P<3ki8o•(IZ/m5n<t@L)
<T8/1)@)Ilg()G8<4<Z/
2.4.3. Ánh xạ giữa 2 loại bảng
m5nm.<Z<T<z<}9r<*<(,f<4<(T<3)([/c(^(
Gi8†(,*([3)SGi8‡(;)(I<=8Xs9G@)0%<oN
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 21
Báo cáo đồ án tốt nghiệp
Chương III. PHÂN TÍCH CÂU HỎI DỰA VÀO TRIE
1. Cấu trúc cây Trie.
c(<M.,)&y(&*@…/&(&(8@†#``\‡y@m/c(<e%(,^<9G@)0%3P<3ki8
od)/c(@%B(30V%.†-‡••†-˜8#‡y†-˜8‡yŒy†-˜8,‡–yX;)-@m/c((B5P5<4<CM%
8@58o&(•’8ž“,ž•#yXm-˜8@m(e(<N<4<<%Š)<=8-ow(3n%X;)8
,*0(1m.y(8'D9p<e%(,^<(,)&X;)<4<‚<%Š)ƒ@m<4<3k9L<M%
2)Xm‚8@58o&(ƒ@m(B5<4<([3S2)Xm<4<9L(H<(Sc(3k9L<M%2)@m
/c(3L)9)I<=8o)S%(T<RKX;)5now(3n%Xmlg((^<3P<349e%y5n<t
@L)3P<3L)9)0o•<4<9L(H<(S
Ví dụ: <Q AT=’NAME’> Who is the
<ENAME type =”POS”> dean</ENAME> of
<ENAME type = “ORG”> Macquarie
University </ENAME> ? </Q>
Có thể được viết lại thành : “^Who is the !POS of !ORG $”. (1)
6)S%9)•<=8<M%2)(,)&(31))1o)S%9)•<=8/c(([3)S(,)&y(%.
)IyX)0<@%(,G(F()y'D9p9e%)0%†(*l&‡Œ@ml4<o)0(,*o)S%9)•
(,)&y/Š)l)/c(^(3P<(z/*q<(L*/;)y(n'%e(K<=8Z3P<)@L)f
/c(^((,*(,)&<Z(S3g3P<V%8)>%9L<M%l4<8%y/Š)^(<Z@%(,G
(n'1Xm5M@*L)v)N'D<Z/c((B5<4<<M%2)y(85M(U<<^(m<4<l%F
9L<p(S'8%ˆ

Bảng 2. Các câu hỏi mẫu
Câu hỏi Khuôn dạng EAT
Y&,&)')@&” Y&,&)'Ÿ"› "
Y*)'(&9&8*:-” Y*)'(&Ÿ?"›*:Ÿ"Rv› K
Y*)' -/)(” Y*)'ŸK› K-
Y*)' -/)(*:-” Y*)'ŸK›*:"Rv› K-
*+:8,)'(&'” *+:8,)'Ÿ"› "
*+(8@@)'-()” *+(8@@)'ŸK› "
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 22
Báo cáo đồ án tốt nghiệp
f9;)<},8<4<9G@)0%3P<(•<T<Xm@%(,G(TX;)(B5<%Š)
3k9L<M%2)doN(,IZ(SB(e.,•y/Š)^(<=8(,)&@%(,G/c(8.
)>%K
Hình 5. Cấu trúc trie xác định thực thể
Bảng 3. Thông tin về các node trong trie
Node (EAT, Tần suất)
# †"y#‡y†Ky#‡y†K-y‡y†6KR‡
7Q †"y#‡
W7 †Ky#‡y†K-y‡
\7# †Ky#‡
# †K-y‡
#J7# †K-y#‡
#\ †6KRy‡
#`7W †6KRy#‡
2. Phân tích câu hỏi dựa vào cấu trúc trie.
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 23
Báo cáo đồ án tốt nghiệp
6;<3n%()I(,*X)0<5M(U<<M%2)y3Z@m<%.S<M%2)X>G3k
9L)1†#‡G3k9LXB.3t)2)349e%5now(3n%Xmlg(
(^<<=8<M%2)yXm<Z(So)g(3P<9L(8.(g<=8/c('1/r%<M%2)9H8Xm*

(H<(S<=8<^
-D9p<4<3k9L<M%2)y(8<1w(f/l)g/'H((U<)G8o)S%9)•
†(*l&‡3n%()I<=8<M%2)X;)^(<=8(,)&g%(f/3P<y(*l&()g5(&*'„3P<(f/
l)g/(,*G^(wX;)^(3n%()I3Z$%4(,fm.@q5@L)<*3gl)lF
<t(*l&m*5N)l)S/(,8*q<(*l&)0(L)lF(S(f/3P<(,*9u.^((,)&
38(f/$%4(,fm.(,NX>KX;)(n'1<8*e(<=8^((,)&<%1)<s3P<
X)g(z/F()m.'„3P<'D9p@m/K<=8<M%2)383P<5M(U<
g%(*l&)0(L)lF(28/uoe(lš^(m*y(*l&()g5(&*'„3P<l)S/
(,8X;)<4<^((,)&(,*9u.f9;)(S)0X)0<(f/l)g/‚o2V%8ƒ(,*V%4
(,f5M(U<<M%2)‚Y*)' *-/)(*:8<V%8,)&)X&,')(.”ƒXm‚Y*)'
*-/)(”ƒ
Hình 6. Duyệt trie
)0<5M(U<<M%2)‚Y*)' *-/)(”ƒ3P<(H<)0o•X)0<l)S/(,8
(*l&/d3n%<M%X;)([‚+*ƒXm‚)'ƒxSE,•([‚ *ƒXm‚-/)(ƒ(,*‚ *
-/)(ƒlF3P<(8.(go•9L(H<(S<=8<^yod)Xf(IlF3P<o)g(
od)v8¡&((&&,[lZ8‚ *ƒlF3P<(f/(e.d^(()g5(&*^(‚)'ƒ†^(#‡yXf
(g([()g5(&*<=8<M%2)†‚-/)(ƒ‡3P<(f/l)g/d^(()g5(&*†^(#Jy#Q‡7
G^(<4<^(ne((f/3P<o;<g%^((f/3P<y(e(<N<4<([lF
(f/(e.d5U8(,;<3>%3P<w<s@*L)X;)<4<^(d)G8^((f/3P<
g%(f/3P<^(5sP5(f<*3OX;)(n'%e(@;'„<)g/%
(g,*V%4(,fm.y<4<^(•/)G8<4<^(5sP53P<(f/(e.'„3k
i89L(H<(S<=8G<p/([lF5sP5(,*9L<M%2)g%(e(<N
<4<(*l&3P<l)S/(,8/mXrlFB9)03P<^(5sP5(f^(<%1)<s
5sP5'„3ki89LK
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 24
Báo cáo đồ án tốt nghiệp
Hình 7. Cấu trúc trie không xác định loại thực thể
,a(M/<=8<M%2)3P<<},8OXm*5n(H<(S<=8(S)09L3k
9L<M%2)-H(8.(g/c(Xm)<p/([(,*<M%2)o•<4<9L(H<(S<Z(S
3P<(H<)0(,;<*q<(,*V%4(,f9%.0((,I<M.(,)&?545m.V%8

(M/3g<NX)0<lF(S(f/(e.'H5sP5)G8(,)&Xm<M%2)g%(,OP53Z
CN.,8yKX;)(n'1<8*e(<=8^(<%1)<s3P<(z/'„3P<4<*<M%2)
U9pˆK<ZC4<'%e(<8*e(d^(W'„3P<4<*<M%2)‚Y*”ƒ
?545(f/l)g/KlFe(()g(.I%<n%X)0<349e%<=8<4<(H<
(S,*(,OP5m.y(,)&3P<CM.9H<}9H8Xm*(F()<T8(,*<4<([
<=8<M%2))<4<(H<(SXm9L(H<(SlF3P<349e%yV%4(,f5M
(U<<M%2)Xr(H<)0()g(,f(f/l)g/o2V%8/F(Nd(,Iy(%.)Iy(,*
(,OP5m.y()g(,flF3ki8o•/c(Z/(H<(S/mo•/1)
V%80<8o)g()G8/c(([(,*CM%X;)/c(([8.<p/([l4<(,*<M%2)38
3P<5M(U<
xS@m/'4(2Xe3>m.yC…(XU9pˆ‚Y*)'(&89/))'(,8()X&8'')'(8(*:
8<V%8,)&)X&,')(.”ƒf<N‚89/))'(,8()X&ƒXm‚8'')'(8(ƒ3>%lF(f/3P<
(,*(,)&df(,I)g(,fCD@E†(f/l)g/o2V%8‡w([‚*:ƒX;)^(#by/q<
3k,•<Z/1)V%80)G8‚89/))'(,8()X&8'')'(8(ƒXZ)‚9&8ƒ(H)G8
‚8<V%8,)&)X&,')(.ƒX;)‚-ƒ
)0<3)([^(m.'8^(l4<<=8(,)&lFe(()g(o•#([y/m<Z(S
o•/c(Z/([3hi8*q<3h<T<zG5453P<3ki8([(,;<
†<Z(SOY*,9&(*q</c(@*L):)@&((H‡
M.(,)&lF<}<Z(4<9p5M(U<<M%2)/m9*3q<3)S/<e%(,^<<M.y
)^5(8<Z(S5M@*L)3P<<4<@*L)<M%2)3S(H<)0(,U<,^((F()'8%m.
)0%V%N<=80(1'„5p(%c<(,H<()g5Xm*'H<UC4<Xm3n.3=<=8<M.
(,)&y(UP5@Ey<UC4<y3n.3=<=8(,)&@L)5p(%c<Xm*lNz3*4By9H
Sinh viên thực hiện: Nguyễn Thúy Ngân – K50 – HTTTB Trang 25

×