可清除樣式集為資料庫中帶來價值較低的樣式集，此研究的應用包含零售業市場分析，以及節省資金等。過往的可清除樣式集演算法，只考慮產品所帶來的利益而沒有考慮到產品中零件各自的權重，且只適用於靜態資料庫中。然而，當新的資料被加入資料庫，原來的資料的門檻值將受到改變，因此必須重新掃描整個資料庫一次。另一方面，考慮產品中各零件的權重及價值探勘分析可清除權重樣式集，以貼近真實生活。但是，當我們使用過往的可清除樣式集演算法並考慮物品的權重，將會違反anti-monotone property。也就是，樣式Y雖然是可清除樣式集，但是其子集X，不一定是可清除樣式集。因此，為了加強過往可清除樣式集的演算法應用範圍，克服上述的限制，許多新的演算法提出了適用於動態資料庫的可清除權重樣式集探勘方法，其中包含IWEI演算法。IWEI演算法使用一個樣式集獲利高估的估計指標，進而滿足anti-monotone property，並使用IWEI-Tree及 OP-List，進而實現適用於動態資料庫演算法。然而，IWEI-Tree在被建立之後，仍必須被重建一次。若資料庫更新的頻率相當頻繁，這將會是一項非常耗時的工作。此外，雖然IWEI演算法使用了一靜態估計指標來刪除可清除權重樣式集之候選集，但由於此靜態估計指標的值仍太大以致於無法有效地刪除候選集。為了解決此問題，於此篇論文，我們提出了一個基於商品之反向索引檔之List結構以及一個區域性的估計指標，並使用商品之反向索引檔之List產生候選集List，再利用此區域性估計指標來刪除候選集List中的候選集，來降低選集List的大小。而我們提出的這個區域性估計指標稱為LMAW，LMAW用來判斷某一樣式集是否有機會成為可清除權重樣式之候選集。我們所提出的一個基於商品之反向索引檔之List演算法，InvP-List演算法，只需要掃描資料庫一次。此外，我們演算法提出的LMAW估計指標，相較於IWEI演算法的估計指標，可產生較少的候選集。從效能結果得知，我們利用真實資料與模擬資料測試，我們提出的InvP-List演算法都比IWEI演算法更有效率。

An erasable itemset is the low profit itemset in the product database, and it is strongly concerned in some applications, including retail market analysis and economized funds. The previous algorithms for mining erasable itemsets ignore the weight of each component of the product and mine erasable itemsets by concerning the product profit only in static product databases. However, while the new data is inserted into the database, the threshold value of the original data will be changed and the database must be scanned again. On the other hand, by considering that each item has its individual price or weight, the result of erasable mining is approaching the real world. But, when we consider the weight of each component, previous algorithms for mining weighted erasable itemsets would violate the anti-monotone property. That is, the subset X of an erasable pattern Y may not be an erasable pattern. Therefore, previous algorithms for mining erasable itemsets need to expand its ability to deal with those kinds of tasks. To deal with the limitation, algorithms for mining weighted erasable itemsets for dynamic databases have been proposed, including the IWEI algorithm. The IWEI algorithm uses the static overestimated factor of itemsets profits to satisfy the ”anti-monotone property” of weighted erasable itemset, and constructs the IWEI-Tree and OP-List data structures for the dynamic database. However, the IWEI-Tree has to be reconstructed, when reading the whole product database is finished. It will take long time to complete the mining of the whole tree, if the database is frequently updated. Moreover, the IWEI algorithm generates the too low static value of the overestimated factor to prune candidates. To solve those problems, in this project, we propose the Inverted-Product-List algorithm (InvP-List) and with the local estimated factor to identify weighted erasable itemsets candidates from the Candidate-List which is generated from InvP-List. We propose the appropriate estimated factor to reduce the number of candidates which is called LMAW. LMAW is a local estimated factor which is used to check whether the itemset is a weighted erasable itemset or not. Our InvP-List algorithm requires only one database scan. Moreover, our proposed algorithm concerning the local estimated factor creates few number of candidates than the IWEI algorithm. From the performance study, we show that our InvP-List algorithm is more efficient than the IWEI algorithm both in the real and the synthetic datasets.

[THESIS VALIDATION LETTER+i]
[ACKNOWLEDGEMENTS+ii]
[ABSTRACT (CHINESE)+iii]
[ABSTRACT (ENGLISH)+iv]
[LIST OF FIGURES+vii]
[LIST OF TABLES+x]
[Introduction+1]
[Weighted Erasable Itemset Mining+2]
[Applications+6]
[Related Works+7]
[Motivation+10]
[Organization of the Thesis+11]
[A Survey of Algorithms for Mining Weighted Erasable itemset+12]
[MERIT Algorithm+12]
[The dMERIT+ Algorithm+13]
[The MEI Algorithm+15]
[The IWEI Algorithm+17]
[Data Structures for the IWEI algorithm+18]
[The Construction of the IWEI-Tree+19]
[The Construction of the OP-list+21]
[The Mining process+21]
[The Inverted-Product List Based Algorithm+23]
[Data Structure+23]
[The Inverted-Product Table+25]
[The Weight Table+31]
[The Profit Table+31]
[The 1-Candidate Table+32]
[The Mining Algorithm with 1-Candidate-Table+34]
[Insertion of the New Data+41]
[Comparison+46]
[Performance+49]
[The Performance Model+49]
[Experiments Results+51]
[Real Dataset+51]
[Synthetic Dataset+56]
[Conclusion+63]
[Summary+63]
[Future Work+64]
[BIBLIOGRAPHY+65]

[1] R. Agrawal and R. Srikant, “Fast Algorithms for Mining Association Rules,”
Proc. of the 20th Int. Conf. on VLDB, pp. 490–501, 1994.
[2] R. Agrawal, T. Imieli´nski, and A. Swami, “Mining Association Rules Between
Sets of Items in Large Databases,” Proc. of the ACM SIGMOD Int. Conf. on
Management of Data, pp. 207–216, 1993.
[3] Z.-H. Deng, G.-D. Fang, Z.-H.Wang, and X.-R. Xu, “Mining Erasable Itemsets,”
Proc. of the Int. Conf.on Machine Learning and Cybernetics, pp. 67–73, 2009.
[4] Z.-H. Deng and X.-R. Xu, “Fast Mining Erasable Itemsets Using NC-sets,” Expert
Systems with Applications, Vol. 39, No. 4, pp. 4453–4463, March 2012.
[5] Z. Deng and X. Xu, “An Efficient Algorithm for Mining Erasable Itemsets,”
Proc. of the Int. Conf. on Advanced Data Mining and Applications, pp. 214–225,
2010.
[6] A. Fariha, C. F. Ahmed, C. K. Leung, M. Samiullah, S. Pervin, and L. Cao,
“A New Framework for Mining Frequent Interaction Patterns from Meeting
Databases,” Engineering Applications of Artificial Intelligence, Vol. 45, pp. 103–
118, Oct. 2015.
[7] J. Han, J. Peid, and Y. Yin, “Mining Frequent Patterns without Candidate
Generation,” Proc. of the ACM SIGMOD Int. Conf. on Management of Data,
pp. 1–12, 2000.
[8] Q. Huynh-Thi-Le, B. V. Tuong Le, and B. Le, “An Efficient and Effective Algorithm
for Mining Top-Rank-k Frequent Patterns,” Expert Systems with Applications,
Vol. 42, No. 1, pp. 156–164, Jan. 2015.
[9] L. J. Kao, Y. P. Huang, and F. E. Sandnes, “Mining Time-Dependent Influential
Users in Facebook Fans Group,” IEEE Int. Conf. on Systems, Man, and
Cybernetics, pp. 718–723, Oct. 2016.
[10] R. U. Kiran, M. Kitsuregawa, and P. K. Reddy, “Efficient Discovery of Periodic-
Frequent Patterns in Very Large Databases,” Journal of Systems and Software,
Vol. 112, pp. 110–121, Feb. 2016.
[11] T. Le, B. Vo, and F. Coenen, “An Efficient Algorithm for Mining Erasable
Itemsets Using the Difference of NC-Sets,” Proc. of the IEEE Int. Conf. on
Systems, Man, and Cybernetics, pp. 2270–2274, 2013.
[12] T. Le and B. Vo, “MEI: An Efficient Algorithm for Mining Erasable Itemsets,”
Engineering Applications of Artificial Intelligence, Vol. 27, No. 1, pp. 155–166,
Jan. 2014.
[13] G. Lee, U. Yun, H. Ryang, and D. Kim, “Erasable Itemset Mining Over Incremental
Databases with Weight Conditions,” Engineering Applications of Artificial
Intelligence, Vol. 52, No. 1, pp. 213–234, June 2016.
[14] G. Lee, U. Yun, and K. H. Ryu, “Sliding Window Based Weighted Maximal
Frequent Pattern Mining Over Data Streams,” Expert Systems with Applications,
Vol. 41, No. 2, pp. 694–708, Feb. 2014.
[15] J. C.-W. Lin, W. Gan, P. Fournier-Viger, and T.-P. Hong, “RWFIM: Recent
Weighted-Frequent Itemsets Mining,” Engineering Applications of Artificial Intelligence,
Vol. 45, No. 1, pp. 18–32, Oct. 2015.
[16] H. Ryang and U. Yun, “High Utility Pattern Mining Over Data Streams with
Sliding Window Technique,” Expert Systems with Applications, Vol. 57, No. 1,
pp. 214–231, Sept. 2016.
[17] J. Sahoo, A. K. Das, and A. Goswami, “An Efficient Approach for Mining Association
Rules from High Utility Itemsets,” Expert Systems with Applications,
Vol. 42, No. 13, pp. 5754 – 5778, Aug. 2015.
[18] C.-H. Weng, “Revenue Prediction by Mining Frequent Itemsets with Customer
Analysis,” Engineering Applications of Artificial Intelligence, Vol. 63, No. 1,
pp. 85–97, Aug. 2017.
[19] U. Yun and G. Lee, “Incremental Mining of Weighted Maximal Frequent Itemsets
from Dynamic Databases,” Expert Systems with Applications, Vol. 54, No. 1,
pp. 304–327, July 2016.
[20] U. Yun and G. Lee, “Sliding Window Based Weighted Erasable Stream Pattern
Mining for Stream Data Applications,” Future Generation Computer Systems,
Vol. 59, No. 1, pp. 1–20, June 2016.
[21] U. Yun, G. Lee, and K. H. Ryu, “Mining Maximal Frequent Patterns by Considering
Weight Conditions Over Data Streams,” Knowledge-Based Systems,
Vol. 55, No. 1, pp. 49–65, Jan. 2014.
[22] U. Yun and H. Ryang, “Incremental High Utility Pattern Mining with Static and
Dynamic Databases,” Applied Intelligence, Vol. 42, No. 2, pp. 323–352, March
2015.
[23] U. Yun, H. Ryang, G. Lee, and H. Fujita, “An Efficient Algorithm for Mining
High Utility Patterns from Incremental Databases with One Database Scan,”
Knowledge-Based Systems, Vol. 124, No. 1, pp. 188–206, May 2017.