Natural Language Interface to Database Access

1
Natural Language Interface to Database Access
Project Supervisor Dr. G. C. Nandi

• Amit Tripathi(19991012)
• Pravi Tripathi(19991035)

2
Purpose Features

• Language Independent

• User Friendly

• Practical Application

• Shifts Burden

• Productivity

• Capabilities

3
Introduction NLP/SQL

• NLP Natural Language Processing
• SQL Structured Query Language
• NLDA Natural Language Database Access

4
Screenshots NLDA_at_IIITA

• VIEW
• QUERY SQL Database
• SQL Command Interface
• OUTPUT

5
Modules

• Collection of Database Info
• Database Interface
• Parser for Smart Dates
• Parsing NL Queries
• SQL queries -gt NL
• User Interface Help

6
Collection of Database Info

• Methods for class DBInfo
• DBInfo
• addTable
• clearTables
• debug
• findColumnName
• isColumn
• isTable

7
Database Interface

• Methods for class DBInterface
• checkConnection
• doInit
• getColumnNames
• resultSetToString
• update

8
Parser for Smartdates

• Detects the presence of legal date string
• Multiple date formats
• JavaCalendar and SimpleDateFormat classes

9
Why Java?

• Platform Independent
• Easy Interfacing with database
• Portable

10
Practical ApplicationEducational/Commercial
11
Modules

• Collection of Database Info
• Database Interface
• Parser for Smart Dates
• Parsing NL Queries
• SQL response --gt NL
• User Interface Help

12
Goals

• Provide Natural Language query Interface
• Provide Response in Natural Language

13
Lexicon

• Show show,list,display,tell
• Noise data,info,information,all,everything
• More more,greater,larger etc..
• Less less,smaller etc..
• Equals equals,contains,is etc..
• Where where
• After after
• Before before

14
Work Flow
Start
Initialization
UI created
Query
Parser
1
Removal Of Level1 Noise
Mode Decision
15
Work Flow contd..
Level2 Noise Removal
1
Eating Words
Filling Structures
Parser for Date Formats
Formation Of SQL query
2
Connection Establishment
16
Work Flow contd..
2
Checking the Connection
Query Submission
ResultSet Structure
Close Connection
3
17
Data Structures Formed

• Database name
• Tables name
• Columns name
• Quantity
• Date
• AND/OR operations

18
Output in KAWA

• query() a_querylist name, phone number, and
  email address of hari
• parse(list name PhoneNumber and Email of hari )
  number of words in sentence7
• searchColumnNameNameTable.name
• query SELECT NameTable.name,
  NameTable.PhoneNumber, NameTable.email FROM
  NameTable WHERE NameTable.name 'hari
• NLEngineLocal.toEnglish(hari,12345678,hari_at_yahoo.c
  om)

19
Work Flow contd..
3
SQL Output
Formatting
Synonym Substitution
End
UI Output
20
Database View through App
21

• Questions/Suggestions
• Thank You