使用子查询改进 JOIN

我捕获对其他用户的文章发表评论的用户，如

INSERT IGNORE INTO AuthorCommentators (AuthorID,CommentatorID,Freq) 
    SELECT AuthorID,CommentatorID,C FROM 
        (SELECT b.UserID AS AuthorID,c.UserID AS CommentatorID,COUNT(*) AS C 
            FROM ArticleComments a  
                JOIN ArticleMap b ON a.ArticleID=b.ArticleID 
                JOIN CommentMap c ON a.CommentID=c.CommentID 
                GROUP BY b.UserID,c.UserID
         ) c 
ON DUPLICATE KEY UPDATE Freq=c.C

但由于ArticleComments表太大，查询速度很慢。

有什么办法可以改进这个查询吗？

桌子很简单，

CREATE TABLE AuthorCommentators
(
AuthorID int(11) unsigned NOT NULL,
CommentatorID int(11) unsigned NOT NULL,
Freq mediumint(7) unsigned,
INDEX(CommentatorID),
INDEX(Freq),
PRIMARY KEY(AuthorID,CommentatorID)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE utf8_general_ci;

CREATE TABLE ArticleComments
(
ArticleID int(11) unsigned NOT NULL,
CommentID int(11) unsigned NOT NULL,
INDEX(CommentID),
PRIMARY KEY(ArticleID,CommentID)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE utf8_general_ci;

CREATE TABLE ArticleMap
(
ArticleID int(11) unsigned NOT NULL,
UserID int(11) unsigned NOT NULL,
INDEX(UserID),
PRIMARY KEY(ArticleID,UserID)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE utf8_general_ci;

CREATE TABLE CommentMap
(
CommentID int(11) unsigned NOT NULL,
UserID int(11) unsigned NOT NULL,
INDEX(UserID),
PRIMARY KEY(CommentID,UserID)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE utf8_general_ci;

由于两者都Articles可以Comments有多个作者，我必须规范化架构。