Application of Data Mining in A Centralized Billing System

Abstract: This article first introduces the concept and related technologies of data mining, then discusses the application of data mining technology in the Centralized Billing System, and uses distributed object technology, multi-layer architecture, Web: the component + B/S + Java + Internet architecture effectively describes the implementation of data mining.
Key words: data mining; Centralized Billing System; JSP/servlet; EJB; B/s;

Abstract: The paper introduces the concept and technology of data mining; then discuss the application of data mining in the concentric billing system; lastly use some tools such as distributed object technology, multi-tier framework and web describe the realization of data mining.

Key words: data mining; concentric billing system; JSP/servlet; EJB; B/s;

1. Introduction:

With the reform of China's telecom industry, operators have been fiercely competitive in fields such as major customers, long-distance businesses, IP businesses, and mobile businesses. In this situation, carriers are most concerned about how to find their most effective customers, how to develop competitive businesses, and how to improve business efficiency. With the rapid development of computer technology, operators have turned to IT technologies to find the best way to win the competition.

The billing and accounting system is closely related to the interests of the majority of telecom users and directly reflects the operator's operating conditions. Therefore, all operators take the construction, maintenance and transformation of the billing and accounting system as the focus of their work, to improve business management efficiency and service quality, major operators have established their own centralized billing systems. It implements unified pricing policies, marketing policies, and service specifications within the billing scope, this service enables centralized online billing, centralized accounting, distribution of business charges, and diversified customer services.

At the same time, the Centralized Billing system also leads to a sharp increase in data, which hides a lot of important information. We hope to analyze the data at a higher level, to make better use of the data. Although the current database system can efficiently implement data input, query, statistics, and other functions, it cannot discover the relationships and rules in the data, the future development trend cannot be predicted based on the existing data, and the hidden knowledge behind data mining is lacking, leading to the phenomenon of "data explosion but poor knowledge.

2. Data Mining Technology Overview:

2.1 Definition of Data Mining:

Simply put, data mining is a massive, incomplete, noisy, fuzzy, and random amount of actual application data, the process of extracting potentially useful information and knowledge hidden in it that people do not know beforehand.

To be precise, data mining is a  in a data warehouse. Under the role of topic-oriented pre-operations, it is converted into a topic-oriented data mining set, then, the data mining set is converted into the corresponding information under the role of the data conversion operation of the mining algorithm. Finally, the information is measured and filtered out when the information is output.

2.2 basic steps for data mining:

First, we need to define business issues and then create a data mining database based on the selected group. The information in the data mining database can be extracted from the Data Warehouse. If you need other information, you can also directly obtain it from the external data source.

Second, after creating a data mining database, you need to analyze the data and develop a preliminary data model, including selecting variables, selecting a record set, transforming variables or creating new variables.

Finally, make a reasonable evaluation of the model. If the model and the actual system are also relatively large, the model needs to be revised again until the model is close to the actual system and then the model is explained, and transfer the information to the management department as an auxiliary decision-making information.

2.3 Data Mining functions:

(1) Automatic prediction of trends and behaviors: Data Mining automatically searches for predictive information in large databases and quickly draws conclusions from the data itself.

(2) Association Analysis: Data Association is an important and discoverable knowledge in the database. If there is a regularity between the values of two or more variables, it is called Association. Associations can be divided into simple associations, time series associations, and causal associations. The purpose of association analysis is to find hidden associated networks in the database. This provides necessary support for certain decisions.

(3) clustering: the records in the database can be divided into a series of meaningful subsets, that is, clustering. Clustering enhances people's understanding of objective reality and is a prerequisite for conceptual description and Deviation Analysis. Clustering technology mainly includes traditional pattern recognition methods and mathematical taxonomy.

(4) concept description: it describes the connotation of a certain object and summarizes the relevant features of such objects. Conceptual descriptions are divided into characteristic descriptions and distinctive descriptions. The former describes the common features of a certain object, and the latter describes the differences between objects.

(5) deviation Detection: the basic method of deviation detection is to find meaningful differences between the observed results and reference values.

2.4 Common technologies for data mining:

Artificial Neural Network: modeled after the non-linear prediction model of the physiological neural network structure, pattern recognition is performed through learning.

Decision tree: represents the Tree Structure of the decision set.

3. Genetic Algorithms: Based on evolutionary theory, they are optimized using methods such as genetic integration, genetic variation, and natural selection.

4. Nearest Neighbor Algorithm: A method used to classify each record in a dataset.

5. Rule derivation: searches for and derives the "if-then" rule in the data in a statistical sense.

2.5 common tools for data mining:

(1) neural network-based tools: due to the rapid modeling capability of non-linear data, neural networks are suitable for non-linear data and noise-containing data, therefore, it is widely used in the analysis and modeling of market databases.

(2) Tools Based on association rules and decision trees: most data mining tools use rule discovery or decision tree classification techniques to discover data patterns and rules. The core of these tools is an inductive algorithm.

(3) tools based on fuzzy logic: the discovery method is to use fuzzy logic for Data Query and sorting.

(4) integrated multi-method tools: many data mining tools use a variety of mining methods. These tools are generally large-scale and suitable for large databases or parallel databases.

3. Application of Data Mining Technology in Centralized Billing System

3.1 overall system structure:

The client only needs one browser, such as IE and Netscape. for near-end customers, you can also develop custom GUI implementations.

The Web server uses JSP servlet to respond to customer needs. The client can access the client through TCP/IP protocol, or directly access the application server to retrieve distributed component objects on the application server.

The application server set is the core part of the system, data mining manager, Data Mining Engine, data converter, data definition, mining wizard, pattern filtering, data preprocessing, mining kernel, pattern expression and interpretation are all Component Objects developed using EJB under the CORBA standard. encapsulate business logic, interfaces for accessing data sources are implemented using JDBC middleware, which can interact with systems such as mining databases and data warehouse data mart. Special interfaces should be developed for file systems.

The overall framework of the system is 3-1:

Figure 3-1

3.2 application instance analysis:

This is an example of analyzing user call characteristics and historical payment records by telecom operators.

By mining the call time, type, and hotspot area information of users, different customers are classified to find the credit rating and consumption mode of each category of customers, develop appropriate marketing strategies for different users and predict the possibility of customer consumption potential, loss analysis, and phone fraud, to provide support for marketing department decision-making.

Call Feature Analysis: Call Feature Analysis analyzes the characteristics of different types of customers on the call. The features include the following indicators:

1. Call, medium call, and short call according to the call time.

2. The types of calls can be divided into intra-city, intra-network, and long-distance calls.

3. The amount of each call can be divided into high calls (for example, 10 yuan or more) and non-high calls.

4. Check the number of calls in different time periods to find out when the call volume is large (for example,-), and when the call volume is small (for example, early morning ). Different charging standards can also regulate the volume of renewals.

5. Users are divided into users who have never been in arrears according to the user's historical payment situation. users who have been in arrears will be paid within January. users who have been in arrears will be paid within March, and users who have been in arrears within June will be maliciously in arrears.

3.2.1 overall design:

Based on the features of each module, the functions of each module are divided into three layers: presentation layer, business logic layer, and data layer. Presentation Layer: This layer mainly completes user-system interaction and simple data processing. Application logic layer: This layer mainly completes complex applications and integrates server components for calling. Database interaction middleware is deployed at this layer, which is part of interaction with data layers. Data level: the DBMS database management system, where data tables and views can encapsulate stored procedures for calling to improve execution efficiency.

3.2.2 presentation layer design:

When a remote browser requests a response on the jsppage of the web server, the user can enter the phone number or contract number of any user in input.html, and then mining. JSP response: the server calls the Mining Component encapsulated by the application server to implement the mining function. The result is returned in HTML format and returned to the browser, at last, we will know the user's credit, consumption potential, and marketing strategies. (Source Code omitted)

3.2.3 logic layer design:

The function we want to achieve is to enter the phone number or contract number of any user, and we will know the user's credit, consumption potential, and marketing strategy. The Core Component can implement functions in one way. It is located on the application server for the Web server to call: In our first example, we need to create a bean entity, a home interface, and a remote interface.

Use JBuilder to create a project named ejbtest (note that the Lib of J2EE should be added to the project ). Then we will add mining. Java (remote interface implementation), mininghome. Java (home interface implementation), and miningejb. Java (EJB object file ). (Source Code omitted)

3.2.4 data layer design:

Store the required data in the corresponding database or data warehouse as the data source for mining and analysis. Because each telecom operator needs its own data warehouse, we will not detail it here.

4. Conclusion

In modern society, the core of most business processes of a company is data. According to statistics, only 7% of the data in a large enterprise database is well applied. In this way, while people feel "excessive data" and "information explosion", they feel "information poor) and "data in jail", and the task of data mining is to find useful data in massive data. However, it is not enough to discover data. We must respond to this model and take actions to convert useful data into information, information into action, and action into value. In the more intense market competition in the future, data mining technology will surely respond faster than others and win more business opportunities.