Association Mining Technology Applied Research In The Sales Of Goods

Data mining is an effective method to find potential and interesting knowledge from mass of data, which is a hot research topic in the field of databases and artificial intelligence. Association rule mining is an important research of data mining and is used to find correlation among item sets from a large amount of data, which has been widely used in e-commerce, merchandising, banking, telecommunication industries and so on.According to the number of abstraction layers related to rules, association rule are divided into single-level association rule and multi-level association rule. Compared with the single-level association rule, multi-level association rule can employ a variety of mining strategies and dig out association rules in one layer or between layers, so it could provide richer and more universal knowledge. In addition, it is very necessary for mining association rules from different abstraction layers. Particularly in e-commerce applications, it is hard to find strong association rules in lower or original layers'data items.Apriori is the most classical association rule mining algorithm. Some of the follow-up algorithms are obtained by improving or expanding of Apriori algorithm. As Apriori generates candidate sets and needs to repeatedly scan the database, therefore it is less efficient. For the shortcomings of Apriori algorithm, scholars have proposed an efficient mining algorithm-FP-growth algorithm, which does not generate candidate items, and only needs to scan the database twice, so it greatly improves the efficiency of frequent item set mining. FP-growth algorithm is widely used in practical applications, but it also has some shortcomings.This paper improves FP-growth algorithm by employing a storage structure based on Hash table, which reduces the time to find items and improves mining efficiency. Based on single level association rule research, the paper studies multi-layer association mining techniques. Concept hierarchy tree is analyzed deeply and its structure is improved. It is used to help to get layers'frequent 1-item sets. Finally, the two investigations are combined together in multi-level association rule mining,which is applied to a practical e-commerce system. It can find some valuable rules among commodities and help to guide corporate management and decision making.