囲碁における定石や将棋における定跡は,主にゲーム序盤で用いられる決まった手順の互角の応酬であり,長い時間をかけて研究・洗練された人智の結晶である.テトリスやぷよぷよなどの落下型パズルゲームの多くでも,展開を有利にするための定石形が存在し利用されている.これらのゲームを囲碁や将棋と比較すると,自分と相手双方に盤が存在し邪魔は間接的にしか行われない一方で,操作対象の与えられ方(俗にツモと呼ばれる)にランダム性があり,状況に応じて用いる定石や配置順を変えていかなければならない難しさがある.本論文では,関連性行列という形で状態と定石を表現する定石形配置法を提案し,これをぷよぷよにおける連鎖の構成に適用することでその有効性を示す.In Go and Shogi, standard patterns and sequence of moves have been developed over the years by human players, mainly in the opening of the game. Such standard patterns, used by expert players to reach a winning position, also exist in tile-matching video games like Tetris and Puyo-Puyo. In tile-matching games, the interaction between the players is only indirect through separate boards, but the randomness of the tiles appearing in the game is a major difficulty, not found in Go or Shogi. In this paper, we propose a tile arrangement method for finding good moves in tile-matching games, through the use of a relevant matrix that represents the current situation and the knowledge of standard patterns. We show the effectiveness of the proposed method by applying it to the construction of chains in Puyo-Puyo. The resulting Puyo-Puyo AI player is significantly stronger.