Pub. Date | : Jan, 2020 |
---|---|
Product Name | : The IUP Journal of Computer Sciences |
Product Type | : Article |
Product Code | : IJCS30120 |
Author Name | : N Umadevi, N Uthra |
Availability | : YES |
Subject/Domain | : Management |
Download Format | : PDF Format |
No. of Pages | : 10 |
Data increases as the usage of computer increases, then the manipulation of data is must to mine the information perfectly. In the jargon of big data, bulk information plays an inevitable role, where a large amount of information creates an issue while extracting the data. To overcome this issue, many algorithms and techniques have been introduced. One such technique is the High Utility Itemset Mining (HUIM) which is an extension of Frequent Pattern Mining (FPM). HUIM was used to mine the customer interoperating databases like departmental store data, stock market data, etc. In the existing research, there were only predictions on positive utility or negative utility. So, the paper proposes both positive and negative utilities, combined and processed by developing an algorithm called High Utility itemset mining with Positive and Negative Utilities (HUPNU). This algorithm infers by deploying opinion mining to recognize the reviews of the users or customers. By considering the reviews and frequencies of the items bought, the high positive and negative utilities are calculated.
High Utility Itemset Mining (HUIM) is an extended zone of Frequent Pattern Mining (FPM). FPM mines the itemsets (Li et al., 2008; Ahmed et al., 2009; Tseng et al., 2010; and Zihayat and An, 2014) in a transactional dataset. FPM checks the transactions that occur more number of times. For example, bread and jam may be bought together mostly, then bread and jam are the frequent items in the itemset. FPM mines these items without mining its price. FPM calculations do not consider the amount and significance (unit benefit, value, hazard, cost and weight, and so on) of the things. An item may occur once or zero time in a transaction. For instance, if a client purchases five breads, ten breads or twenty breads, it is seen as equivalent. In this manner, market and retailers are keen on finding the higher beneficial itemsets instead of incessant itemsets. To beat this issue, HUI mining techniques were introduced (Liu et al., 2005). In HUI mining, things can have the amount and relative significance. Association rule mining has been broadly considered throughout the decade (Yen and Lee, 2007).
High Utility Itemset Mining (HUIM), Frequent Pattern Mining (FPM), EFIM algorithm, Opinion mining, Association rule mining