TEMPLATE RELATION(TR) RECOGNITION

4. Template Relation(TR) recognition: Find-

tic information is the loss of the robustness caused

ing the relation between TEs and a question

by the error of the NLP components. However,

focus based on TR rules, and syntactic infor-

our IE-based approach for question analysis uses

mation, etc.

the very restricted and essential contextual infor-

mation in each step and can avoid such a risk suc-

Finally, the question analysis component selects

cessfully.

the proper frame for the question and fills proper

values of each slot of the selected frame.

The example of the analysis process of this

Question : “오늘

NBC에서

저녁에

어떤

프로

하니?”

TE Recognition

(today) (on NBC)

(at night)

(what)

(program)(play)

(“What movie will be played on NBC tonight?”

in English)

a

(3) :

Natural Language Analysis

“오늘”/NE_Date

(at night) Question focus : QF_program(today)

“NBC”/NE_Channel

(on NBC)

“저녁”/NE_Time

TE_B

EGIN

D

ATE

TE_C

HANNEL

TE_B

EGIN

Time

(1) :

TR Recognition

(at night)

“어떤”/C_what

(play)(program)

“하니”/C_play

(what)

“프로”/C_prog

REL_OK

Question Focus Recognition

(4) :

(2) :

Question focus : QF_program(at night)

Translation of Semantic Frame

Question focus region Question focus : QF_program

FRM : PROGRAM_QUESTION

Question focus : QF_program

‘NE_*’ denotes that the corresponding word is named entity of *.

Begin Date :

“

Today

”

‘C_*’ denotes that the corresponding word is belong to the concept C_* in the ontology.

Begin Time :

“

Night

”

‘TE_*’ denotes that the corresponding word is template element whose type is *.

Channel :

“

NBC

”

‘REL_OK’ indicates that the corresponding TE and question focus are related.

Figure 2: Example of Question Analysis Process in K-QARD

component is shown in Figure 2-(1).

first part that recognizes the question focus region

is domain-independent.