基于频繁模式树的分布式关联规则挖掘算法-控制与决策.PDFVIP

绗?7 鍗 绗? 鏈 鎺 鍒 涓 鍐 绛 2012 骞 4 鏈 Vol. 27 No. 4 Control and Decision Apr. 2012 鏂囩珷缂栧彿: 1001-0920 (2012) 04-0618-05 鍩轰簬棰戠箒妯″紡鏍戠殑鍒嗗竷寮忓叧鑱旇鍒欐寲鎺樼畻娉 浣 娉 (閲嶅簡鐞嗗伐澶у璁＄畻鏈虹瀛︿笌宸ョ▼瀛﹂櫌锛岄噸搴?00054) 鎽 瑕? 鎻愬嚭涓€绉嶅熀浜庨绻佹ā寮忔爲鐨勫垎甯冨紡鍏宠仈瑙勫垯鎸栨帢绠楁硶(DMARF). DMARF 绠楁硶璁剧疆浜嗕腑蹇冪粨鐐? 鍒╃敤灞€閮 棰戠箒妯″紡鏍戣鍚勮绠楁満缁撶偣蹇€熻幏鍙栧眬閮ㄩ绻侀」闆? 鐒跺悗涓庝腑蹇冪粨鐐逛氦浜掑疄鐜版暟鎹眹鎬? 鏈€缁堣幏寰楀叏灞€棰戠箒椤归泦. DMARF 绠楁硶閲囩敤椤堕儴鍜屽簳閮ㄧ瓥鐣? 鑳藉ぇ骞呭噺灏戝€欓€夐」闆? 闄嶄綆閫氫俊閲? 鐞嗚鍒嗘瀽鍜屽疄楠岀粨鏋滃潎琛ㄦ槑浜咲MARF 绠楁硶 鏄揩閫熻€屾湁鏁堢殑. 鍏抽敭璇岤 鏁版嵁鎸栨帢锛涢绻佹ā寮忔爲锛涘叏灞€棰戠箒椤归泦锛涘叧鑱旇鍒 涓浘鍒嗙被鍙耳 TP311 鏂囩尞鏍囪瘑鐮? A Distributed algorithm for mining association rules based on FP-tree HE Bo (School of Computer Science and Engineering锛孋hongqing University of Technology锛孋hongqing 400054锛孋hina. E-mail 锛歨ebo@cqut.edu.cn) Abstract 锛氾細锛歍he paper proposes a distributed algorithm for mining association rules based on frequent pattern tree(FP-tree), named distributed algorithm for mining association rules based on FP-tree(DMARF) algorithm, which sets center node. DMARF algorithm makes computer nodes compute local frequent itemsets independently from the local FP-tree. Then the center node exchanges data with other computer nodes. Finally, global frequent itemsets are gained. DMARF can decrease candidate itemsets and communication traf铿乧 by using the strategies of top and bottom. Theoretical analysis and experimental results show that DMARF algorithm is fast and effective. Key words 锛氾細锛歞ata mining 锛沠requent pattern tree 锛沢lobal frequent itemsets 锛沘ssociation rules 1 寮曞紩寮 瑷€瑷€瑷€ 绛夐棶棰? 閽堝杩欎簺闂, 鏈枃鎻愬嚭涓€绉嶅熀浜庨绻佹ā寮 鏁版嵁鎸栨帢鏄粠鏁版嵁闆嗕腑鍙戠幇鏂伴鐨勩€侀鍏堟湭鐭 鏍戠殑鍒嗗竷寮忓叧鑱旇鍒欐寲鎺樼畻娉?DMARF). DMARF 绠 鐨勩€侀殣钘忓拰鏈夎叮鐨勭煡璇? 鏁版嵁鎸栨帢涓殑鍏宠仈瑙勫垯鏄 娉曡缃簡涓績缁撶偣, 鍒╃敤灞€閮ㄩ绻