In Part 1 of Semantic Search Implementation that I published yesterday, I wrote about how to install and configure semantic search in SQL Server 2012. Today in Part 2, we will learn how to implement semantic search and will demonstrate how beneficial it is.
Prerequisite :
Given below are the two articles that you should read to have clear understanding about Semantic Search Implementation.
- Install & Configure Semantic Search
- File Table
Lets Implement it step by step.
Step 1: Create database
USE master
GO
CREATE DATABASE [SampleDB]
ON PRIMARY
( NAME = N'SampleDB',
FILENAME = N'C:\DATA\SampleDB.mdf'
),
FILEGROUP SampleFileGroup CONTAINS FILESTREAM
(NAME = SampleFileTableFS,
FILENAME='C:\Data\SampleFileTable'
)
LOG ON
( NAME = N'SampleDB_log',
FILENAME = N'C:\DATA\SampleDB_log.ldf'
)
WITH FILESTREAM
( NON_TRANSACTED_ACCESS = FULL,
DIRECTORY_NAME = N'SampleFileTableDB'
)
GO
Step 2: Create FileTable
USE SampleDB
GO
CREATE TABLE CVBank AS FileTable
WITH
(
FileTable_Directory = 'CVFileTable_Dir'
);
GO
Step 3: Populate data into FileTable
In this step, we will insert few Cvs in the file table.
USE SampleDB
GO
INSERT INTO [dbo].[CVBank] ([name],[file_stream])
SELECT'Muhammad Imran -CV.doc',
* FROM OPENROWSET
(
BULK N'C:\Cv\Muhammad Imran -CV.doc', SINGLE_BLOB
) AS FileData
GO
INSERT INTO [dbo].[CVBank] ([name],[file_stream])
SELECT 'Salman Tahir -CV.doc',
* FROM OPENROWSET
(
BULK N'C:\Cv\Salman Tahir -CV.doc', SINGLE_BLOB
) AS FileData
GO
INSERT INTO [dbo].[CVBank] ([name],[file_stream])
SELECT 'Syed Kamran Ali -CV.doc',
* FROM OPENROWSET
(
BULK N'C:\Cv\Syed Kamran Ali -CV.doc', SINGLE_BLOB
) AS FileData
GO
Step 4: Create full text search on FileTable
Right click on the file table point to full-Text index then select Define Full-Text Index… Further details to create Full Text Index is available here
Now, we are prepared to use semantic search. Lets create few examples.
Example 1: Search Sharepoint Developer CVs
As I mentioned above that, I populated some CVs in the file table, now it is time to search Cvs with certain criteria using semantic search.
In this example, I will search candidates having knowledge of sharepoint.
USE SampleDB
GO
SELECT
D.name
,KEYP_TBL.score
,KEYP_TBL.Keyphrase
FROM dbo.CVBank D
INNER JOIN SEMANTICKEYPHRASETABLE
(
dbo.CVBank,(name,file_stream)
) AS KEYP_TBL
ON D.path_locator = KEYP_TBL.document_key
WHERE KEYP_TBL.keyphrase like '%sharepoint%'
ORDER BY KEYP_TBL.score DESC
--OUTPUT
Example 2: Search SQL Developer CVs
In this example, I will search candidates having knowledge of SQL.
USE SampleDB
GO
SELECT
D.name
,KEYP_TBL.score
,KEYP_TBL.Keyphrase
FROM dbo.CVBank D
INNER JOIN SEMANTICKEYPHRASETABLE
(
dbo.CVBank,(name,file_stream)
) AS KEYP_TBL
ON D.path_locator = KEYP_TBL.document_key
WHERE KEYP_TBL.keyphrase like '%sql%'
ORDER BY KEYP_TBL.score DESC
--OUTPUT
Conclusion :
Semantic Search is very handy if you need to develop some solution on the basis of unstructured document. Shall post another nice feature of semantic search in my upcoming post.
SQL Server Portal 