Agenda for Presenation
Download
Report
Transcript Agenda for Presenation
Agenda for Presenation
What is NLIDB
What has been done
What is to be done
What is NLIDB?
Organized Data
Questions and Answers
Use of SQL
Basically, we are trying to extract
data present in SQL databases
What is the BIG idea?
Correspondence between
relationships and verbs (or
Adjectives)
Concept of verb frames to represent
the relationships
What is a verb frame???
A frame with a central verb and some
arguments
How does this idea work?
Some information we need to provide
to the NLIDB
Understanding the database
The different verb frames present
How this information is presented?
The 5 files that the developer needs to
make
ER representation
Example Format
%er_rep = {
student => {
type => 'entity' ,
attribute => [
Student_ID,Name,Date_OF_Birth,Email,Sex,Address,City,State,Pin,Sex,
CGPA,Program_
Name,Year_Of_Admission ] ,
primary_key => 'Student_ID' ,
rel_ship => 'register'
},
registration => {
type => 'relationship' ,
attribute => [
Student_ID,Course_Code,Course_Grade ],
primary_key => 'Student_ID,Course_Code' ,
entity => 'student,course_offering'
}
Verb Frames
Example Format
NLexpr=STUDENT$ GET GRADE$ IN COURSE$
ERexpr=register(student._key_,course_offering.gra
de,course_offering._key_)
NL-Er_Mapping=
Verb::GET;
SUBJ::STUDENT$::student._key_;
OBJ::GRADE$::registration.Course_Grade;
PP_IN::COURSE$::course_offering.Course_Name;
Definitions
Example Format
STUDENT$=pupil,student;
GRADE$=grade,marks,percentage;
Join Information
Example Format
faculty:course_offering=>FACULTY.Faculty_
ID=COURSE_OFFERING.Faculty_ID
Course_offering:registration=>COURSE_OF
FERING.Course_Code=REGISTRATION.C
ourse_Code
registration:student=>STUDENT.Student_I
D=REGISTRATION.Student_ID
Database Information
Example Format:
IP Address of Database Server =
172.16.9.26
Database Name = academics
UserName = root
Password = nlidb123
What all can the system do?
Understand simple sentences
Understand alternatives of nouns
used
Understand alternatives of verbs used
Come up with a probable answer
What can it be made to do?
A lot of things
Handle complex sentences (Nested
Queries)
Simple Dialog Modeling (A project on
Semantic Completion is done)
Improved Adjectives Handling
Non NLP stuff which can be done
Distribution mechanism for different
domain/database data
Improving database search
Different DBMS
The End
Questions?