如何使用 MS SQL 根据标签、类别、国家对新闻进行分组

Question

我有一个与新闻相关的表格，其中包含多个字段，例如Title, Category, Tag, Country, details ...

我需要根据用户现在正在阅读的新闻显示related news。

相关新闻可以基于Tag, Category & Country

我面临的问题是Tags 我需要使用所有新闻项目搜索每个标签，例如，如果新闻项目有 3 个标签，那么它必须搜索所有新闻记录三次才能获得基于标签的结果匹配。

我在 Fiddle http://sqlfiddle.com/#!18/d9dd7/1 上设置了示例表结构

示例表

CREATE TABLE [TestTable](
    [AutoID] [int] IDENTITY(1,1) NOT NULL,
    [Title] [nvarchar](100) NULL,
    [Category] [nvarchar](50) NULL,
    [Tags] [nvarchar](200) NULL,
    Country [nvarchar](50)
) 

示例

假设我正在阅读This is News One America

然后相关文章应显示This is News five，因为它与类别和两个关键字“唐纳德特朗普，伊朗”匹配，但在这种情况下，国家不同，这有点困难

所以搜索应该优先基于

1. 1 类
2. 标签
3. 国家

我正在使用的 SP

DECLARE @Category NVARCHAR(100);
DECLARE @Country NVARCHAR(100);
DECLARE @Tags NVARCHAR(4000);


    SELECT 
    ,@Category = 'Politics'
    ,@Country = 'US'
    ,@Tags = 'Donald Trump, Iran'
    FROM TestTable WHERE AutoID=1

    DECLARE @tempTable TABLE
    (
        TempNewsId INT IDENTITY(1,1),
        ID INT,
        Title NVARCHAR(MAX),
        Desc NVARCHAR(MAX),
        Tags NVARCHAR(100),
        Country NVARCHAR(50),
        Category NVARCHAR(100)
    )

    IF @Category IS NOT NULL
        BEGIN
            SET @Category ='%'+ @Category+'%'
            INSERT INTO @tempTable
            SELECT TOP 4
            AutoID, Title, Desc,Tags,Country,Category FROM TestTable 
            WHERE Category IS NOT NULL AND Category LIKE @Category
        END 

    IF @Tags IS NOT NULL
        BEGIN
            SET @Tags ='%'+ @Tags+'%'
            INSERT INTO @tempTable
            SELECT TOP 4
            AutoID, Title, Desc,Tags,Country,Category FROM TestTable 
            WHERE Tags IS NOT NULL AND Category LIKE @Tags
        END 




    IF @Country IS NOT NULL
        BEGIN
            SET @Country ='%'+ @Country+'%'
            INSERT INTO @tempTable
            SELECT TOP 4
            AutoID, Title, Desc,Tags,Country,Category FROM TestTable 
            WHERE Country IS NOT NULL AND Country LIKE @Country
        END 

    SELECT TOP 20 ID, Title,[Desc],Tags,Country,Category    FROM @tempTable 
    ORDER BY TempNewsId ASC

Answer 1

性能不会很好，但你可以创建一个标量函数来比较标签列表并返回相似标签的数量。

create FUNCTION dbo.CompareTags
(
    @inputTagList1 varchar(255),
    @inputTagList2 varchar(255),
    @SplitOn varchar(5) = ','
)
RETURNS int
BEGIN

    declare @tagList1 table (Tag varchar(100));
    declare @tagList2 table (Tag varchar(100));

    -- populate table with first list of tags
    DECLARE @split_on_len INT = LEN(@SplitOn)
    DECLARE @start_at INT = 1
    DECLARE @end_at INT
    DECLARE @data_len INT
    WHILE 1=1
    BEGIN
        SET @end_at = CHARINDEX(@SplitOn,@inputTagList1,@start_at)
        SET @data_len = CASE @end_at WHEN 0 THEN LEN(@inputTagList1) ELSE @end_at-@start_at END
        INSERT INTO @tagList1 (Tag) VALUES( lower(rtrim(ltrim(SUBSTRING(@inputTagList1,@start_at,@data_len)))) );
        IF @end_at = 0 BREAK;
        SET @start_at = @end_at + @split_on_len
    END

    -- populate table with second list of tags
    set @start_at = 1
    WHILE 1=1
    BEGIN
        SET @end_at = CHARINDEX(@SplitOn,@inputTagList2,@start_at)
        SET @data_len = CASE @end_at WHEN 0 THEN LEN(@inputTagList2) ELSE @end_at-@start_at END
        INSERT INTO @tagList2 (Tag) VALUES( lower(rtrim(ltrim(SUBSTRING(@inputTagList2,@start_at,@data_len)))) );
        IF @end_at = 0 BREAK;
        SET @start_at = @end_at + @split_on_len
    END


    -- compare tables
    RETURN (select count(*) from @tagList1 as a inner join @tagList2 as b on a.Tag = b.Tag)
END
go