28-06-2013, 03:09 PM
Mining Weighted Association Rules without Preassigned Weights
Mining Weighted Association.docx (Size: 20.37 KB / Downloads: 16)
ABSTRACT
Association rule mining aims to explore large transaction databases for association rules, which may reveal the implicit relationships among the data attributes. It has turned into a thriving research topic in data mining and has numerous practical applications, including cross marketing, classification, text mining, Web log analysis, and recommendation systems.
The classical model of association rule mining employs the support measure, which treats every transaction equally. In contrast, different transactions have different weights in Real-life data sets. For example, in the market basket data, each transaction is recorded with some profit. Much effort has been dedicated to association rule mining with reassigned weights. However, most data types do not come with such preassigned weights, such as Web site click-stream data. There should be some notion of importance in those data. For instance, transactions with a large amount of items should be considered more important than transactions with only one item. Current methods, though, are not able to estimate this type of importance and adjust the mining results by emphasizing the important transactions.
In this paper, we introduce w-support, a new measure of item sets in databases with only binary attributes. The basic idea behind w-support is that a frequent item set may not be as important as it appears, because the weights of transactions are different. These weights are completely derived from the internal structure of the database based on the assumption that good transactions consist of good items.
This assumption is exploited by extending Kleinberg’s HITS model and algorithm to bipartite graphs. Therefore, support is distinct from weighted support in weighted association rule mining (WARM), where item weights are assigned. Furthermore, a new measurement framework of association rules based on w-support is proposed. Experimental results show that w-support can be worked out without much overhead, and interesting patterns may be discovered through this new measurement. The rest of this paper is organized as follows: First, WARM is discussed. Next, we present the evaluation of transactions with HITS, followed by the definition of w-support and the corresponding mining algorithm. An interesting real-life example and experimental results on different types of data are given. Concluding remarks are made in the last section.
EXISTING SYSTEM:
In this system normally users can rate the products in the website .here the top ten products are getting by using pre assigned weights, so apart from top ten products we can display the top one product according to the support system.
PROPOSED SYSTEM:
In this system once a registered user rate a product, analysis is made on the database and the role of the user in the rating system is identified and rate value of the product is concluded depending upon the his role, rate value is taken from the threshold value which we get from analysis conclusion, while the top product is to be calculated a scan is made on the database top ten user are identified then the top product rated by them is found and saved to location, this process is carried out for each and everyone in the top ten user list, finally a analysis is made over top product rated by the top ten user and we get to conclusion which is the best product, by this we get perfect result from this analysis.