Download Android App

Alternate Blog View: Timeslide Sidebar Magazine

Friday, January 13, 2012

Web Usage Mining - Part 1


Web mining is the use of data mining techniques to automatically discover and extract information from Web documents/services.

Web mining is categorized into 3 types.
  1. Content Mining (Examines the content of web pages as well as results of web Searching)
  2. Structure Mining (Exploiting Hyperlink Structure) 
  3. Usage Mining (analyzing user web navigation)
Web usage mining is one of the prominent research area due to these following reasons:
  • One can keep track of previously accessed pages of a user. These pages can be used to identify the typical behavior of the user and to make prediction about desired pages. Thus personalization for a user can be achieved through web usage mining. 
  • Frequent access behavior for the users can be used to identify needed links to improve the overall performance of future accesses. Prefetching and caching policies can be made on the basis of frequently accessed pages to improve latency time. 
  • Common access behaviors of the users can be used to improve the actual design of web pages and for making other modifications to a Web site. 
  • Usage patterns can be used for business intelligence in order to improve sales and advertisement by providing product recommendations.
There are two classes of data mining namely i) to summarize or characterize general properties of data in repository which is called Descriptive and ii) to perform inference on current data, to make predictions based on the historical data which is called Prescriptive. There are various data mining techniques available which also can be applied to web data mining. Few techniques are listed below.

1) Association Rules Mining: When the book Data Mining Concepts and Techniques is bought, 40% of the time the book Database System is bought together, and 25% of the time the book Data Warehouse is bought together. Those rules discovered from the transaction database of the book store can be used to rearrange the way of how to place those related books, which can further make those rules more strong.

2) Sequential Pattern Mining: Association rule mining does not take the time stamp into account, the rule can be Buy A=>Buy B. If we take time stamp into account then we can get more accurate and useful rules such as: Buy A implies Buy B within a week, or usually people Buy A every week. As we can see with the second kind of rules, business organizations can make more accurate and useful prediction and consequently make more sound decisions. A database consists of sequences of values or events that change with
time, is called a time-series database, a time-series database records the valid time of each dataset. For example, in a time-series database that records the sales transaction of a supermarket, each transaction includes an extra attribute indicate when the transaction happened. Timeseries database is widely used to store historical data in a diversity of areas such as, financial data, medical data, scientific data and so on. Different mining techniques have been designed for mining time-series data, basically there are four kinds of patterns we can get from various types of timeseries data:1) Trend analysis, 2) Similarity search, 3) Sequential patterns and 4) Periodical patterns.

Sequential patterns: sequential pattern mining is trying to find the relationships between occurrences of sequential events, to find if there exists any specific order of the occurrences. We can find the sequential patterns of specific individual items; also we can find the sequential patterns cross different items. Sequential pattern mining is widely used in analyzing of DNA sequence. An example of sequential patterns is that every time Microsoft stock drops 5%, IBM stock will also drops at least 4% within three days.

3) Classification: Classification is to build (automatically) a model that can classify a class of objects so as to predict the classification or missing attribute value of future objects (whose class may not be known). It is a two-step process. In the first process, based on the collection of training data set, a model is constructed to describe the characteristics of a set of data classes or concepts. Since data classes or concepts are predefined, this step is also known as supervised learning (i.e., which class the training sample belongs to is provided). In the second step, the model is used to predict the classes of future objects or data. A decision tree for the class of buy laptop, indicate whether or not a customer is likely to purchase a laptop. Each internal node represents a decision based on the value of corresponding attribute, also each leaf node represents a class (the value of buy laptop=Yes or No). After this model of buy laptop has been built, we can predict the likelihood of buying laptop based on a new customer's attributes such as age, degree and profession. That information can be used to target customers of certain products or services, especially widely used in insurance and banking.

4) Clustering: Classification can be taken as supervised learning process, clustering is another mining technique similar to classification. However clustering is a unsupervised learning process. Clustering is the process of grouping a set of physical or abstract objects into classes of similar objects, so that objects within the same cluster must be similar to some extent, also they should be dissimilar to those objects in other clusters. In classification which record belongs which class is predefined, while in clustering there is no predefined classes. In clustering, objects are grouped together based on their similarities. Similarities between objects are defined by similarity functions, usually similarities are quantitatively specified as distance or other measures by corresponding domain experts. For example, based on the expense, deposit and draw patterns of the customers, a bank can clustering the market into different groups of people. For different groups of market, the bank can provide different kinds of loans for houses or cars with different budget plans. In this case the bank can provide a better service, and also make sure that all the loans can be reclaimed.

To be continued in Part 2.

No comments:

Post a Comment