Business Intelligence and Its Implementation Model

1 Introduction

As a brand-new communication media, Internet brings new opportunities in many fields. Web is becoming an important channel for business promotion, and e-commerce is experiencing rapid development. In general, the information or data involved in commerce is very complicated. To handle a large number of e-commerce business data, a certain amount of advanced information processing technology is required. Business Intelligence is an ideal

.

As a tool to assist in business operations and decision-making, business intelligence aims to improve operation efficiency, strive for maximum profit, maintain customer loyalty, manage seasonal variables, and form personalized services. bringing business intelligence tools (such as data mining) into the eyes of e-commerce can reveal potential business opportunities behind data, this has gained popularity among e-businesses. The fundamental purpose of business intelligence is to "make better decisions faster ".

2 business intelligence application scenarios

Based on the characteristics of e-commerce and the advantages of business intelligence, business intelligence can play an important role in the following e-commerce fields.

(1) Customer Relationship Management

Understanding customer behavior helps improve relationships with customers. make correct business decisions. this is because customer relationships can bring the following benefits to enterprises: 1) reduce interaction costs; 2) Transfer work to customers; 3) Monitor company performance; 4) cultivate loyal customers; 5) obtain customer information; 6) Procedural company's business process.

The most important contribution of Internet to the company is that it has the ability to establish one-to-one relationship between the company and customers, thus changing the relationship between the entire customer and making the company more closely and easily connected with end customers. the business challenge is to find and understand the customer's demand, and then provide the right products to meet the customer's requirements. business Intelligence is suitable for doing this kind of work, and can bring more profits and competitive advantages by cultivating customer loyalty and providing special services.

(2) Commercial Negotiation

Negotiation means information exchange between buyers and sellers. commercial negotiation may only focus on price or wider product attributes, product options (such as quality assurance, delivery time, payment method, terms of service, etc. Because the participants are separated from each other, and they may have different business practices and different cultural backgrounds, negotiation in e-commerce is more important and complex. software agents based on business intelligence can participate in business negotiation and overcome many difficulties. The working principle is that the buyer communicates with a software agent, the agent automatically executes the buyer's requirements and provides the seller with the necessary information to complete all the steps of the established Value Chain Model A for negotiation and decision analysis. Software Agents are personalized, autonomous, and adaptive, and their operation modes are preset.

(3) Information security assurance

E-commerce involves a large amount of customer data and sensitive data. The parties involved in the transaction need to perform identity authentication and other operations to ensure the security of e-commerce transactions. access control, confidentiality, integrity, non-repudiation, and other security services are required, the entire transaction activity will involve the implementation of Multiple Information Security Protection behaviors. The commercial principle-based security proxy mechanism can well adapt to and meet such needs.

(4) Logistics Planning

The basic requirements of modern logistics include: real-time tracking of the entire process of logistics operations, such as real-time reporting to the customer about the location of their items, accurately predict the arrival time of an item and notify the customer in a timely manner. The customer can query the transportation and delivery status information of the item online. The logistics operation cost is as low as possible, the response speed of logistics operations should be as timely as possible, and 3rd party logistics management should be accepted. routing planning, vehicle scheduling, and shipping planning in the logistics system must meet these strict requirements. This is a complex task. It is a multi-objective decision-making question that takes the time limit and cost as the main decision factors, to solve such problems, intelligent and efficient decision-making support systems are necessary.

3. Business Intelligence Implementation Model

3.1 identify business intelligence opportunities

The first task to start business intelligence is to clearly define the goals to be achieved, which means to seek opportunities for business intelligence within the Organization to improve the quality of daily decision-making. an easy-to-use process for evaluating business intelligence opportunities within a specific organization can be divided into three basic steps:

1) Collect information: answers questions about "who, what, where, why, when, and how to do" in business activities. think about the possible application fields, beneficiaries, and required information types of business intelligence in an organization;

2) Sharing and gathering ideas: bringing together a group of people to hold brainstorming activities to share with each other "what kind of business process can benefit from business intelligence ?", "What information can help improve these processes ?" Other ideas and experiences;

3) Evaluate the idea: Evaluate the ideas and ideas collected in brainstorming with standards, and identify the Business Intelligence opportunities that can provide the maximum benefit once brainstorming is completed. you can get a list of business intelligence opportunities, group them, evaluate them by importance, and finally get sorted opportunities.

3.2 Key Technologies of business intelligence

3.2.1 Data Mining

Business operations are not isolated. The key to special services is to make full use of accumulated business data. historical data can provide valuable guidance for current decisions. It can help people execute optimized business operations through core businesses, increase market share, and cultivate customer loyalty. however, it has been difficult to effectively reveal the useful modes behind a large amount of data because the data is constantly growing. An effective solution is data mining, which can discover this hidden model and use it to guide future business. data mining is to obtain the behavior patterns hidden in a large amount of data, extract valid, practical, unknown, and complex information, and apply it to business decisions. as a powerful business intelligence tool, data mining is becoming popular. It shortens the distance between data collection and utilization and is suitable for e-commerce of different scales. Data mining can help e-commerce operators improve their customer relationships, make correct decisions, and enhance their competitiveness. data Mining mainly involves the following types of operations: creating a prediction model (using examples to predict the value of an attribute) and database segmentation (using attributes to group records, each group of records has similar properties, but the difference between groups is obvious); associativity analysis (identifying records or time in transactions)

(Identify the database contains records or records that do not expect value). Generally, data mining can achieve the following functions:

(1) statistical functions

The statistical function can help you analyze and forecast data. Statistical functions are divided into: 1) subanalysis-finding relationships between multiple hidden variables; 2) linear attenuation-used to determine the linear relationship between the primary and non-primary variables; 3) main variable analysis-used to convert the coordinate system for better matching of the coordinate axis and data distribution; 4) fitting the heavy variable curve-finding a mathematical function that can accurately describe the data distribution; 5) variable statistics-detailed statistics.

(2) Mining Functions

Mining features include: Correlation Analysis: Correlation Algorithms seek a pattern such as whether to buy a paint brush when buying paint, that is, it determines the probability, for example, if you buy paint, there is a 20% probability that a brush will also be purchased, and thousands of rules will be created when the algorithm is run. You can select a subset of these rules. the credibility of these rules depends on the user's understanding and choice of the rules, and the correlation analysis is used for market basket analysis, promotion plans, and so on. 2) sequence mode: the generation method of rules depends on the correlation algorithm. This algorithm is used to view the series purchase records of a customer. For example, in January, a lunch box and a tent were purchased, and in February, a travel backpack and a videotape were purchased. In March, a sleeping bag was purchased. Here, the Sequence Correlation Algorithm checks all records in the T "table and returns the following rule: if the January purchase target contains a lunch box, the probability of purchasing a sleeping bag in March is 30% sequence mode, and time correlation can be found. 3) clustering: Clustering is used to segment A database into a subset. Members of each cluster have similar attributes. clustering can be achieved through statistics or neural network algorithms, depending on the type of input data. for example, the purchase record can reflect the buyer's model and their preferences for different products. classification (segmentation) of a person based on his/her behavior. Then, observe these segments and identify them using statistics to obtain the correlation between products. the result of the above steps is that a retailer can have a better understanding of different types of customers, and then adjust the market strategy accordingly to meet the different needs of each type of customers, provide appropriate products to them through advertising media. clustering is well used for cross-selling. for personalized marketing services of different customers, it determines the media promotion solution and understanding the shopping target. 4) classification: classification is the process of automatically creating a class model from the record set. A compelling model is composed of multiple patterns. Once a model is introduced, records that help classify data can be used to predict other unclassified records. Available methods include tree extension and neural network reverse propagation. 5) Forecast: Like classification, the purpose of the forecast is to establish a data model with a general record. the difference between them is that the goal is not a class member relationship, but a continuous value or sorting forecast can adopt the neural network algorithm and the radial basis function (RBF) algorithm.

Similarity time series: it aims to discover the probability of occurrence of similar sequences in a time series database. given a time series database, the goal is to find a sequence similar to a given value, or to find the probability of occurrence of all similar sequences.

3.2.2 Neural Networks

In a rapidly changing and fiercely competitive business environment, faster and better decision-making is an important way to win. Market Decision makers are increasingly keen on using computer decision support systems to help them make the right choices. The network has been widely used in the commercial field. Especially when the problem domain involves classification, recognition, and prediction. neural Networks enable the discovery of invisible potential trends and relationships between data. article 2 5 2 explores the application of neural networks to assist decision-making support in retail and B2B e-commerce. The goal is to capture the marketing techniques such as advertising) the complex relationship with the total sales volume is used to find out the ing relationship between output fluctuations caused by the change of input volume, and the important impact factors may be identified through the prediction model and sensitivity analysis of neural networks, this model can achieve good performance in a given short-term forecast.

The use of neural networks can be divided into reverse propagation and Forward propagation. Reverse propagation neural networks are suitable for daily or weekly data forecasts. Compared with reverse propagation, Forward Propagation Neural Networks have more advantages in speed. The Forward propagation neural network is not suitable for daily prediction, but provides good results for weekly prediction. this may be due to the existence of a high correlation between the number of people input in the weekly prediction model. Positive neural networks cannot be used for sensitivity analysis, the market significance of sensitivity analysis conducted by a properly trained neural network model is that the research results can be used in practical application of market management. the results show that:

3.2.3 intelligent proxy

Before making decisions on traditional shopping activities, you need to invest a large amount of manpower to collect and sort out information about related businesses, products and services. it is not easy to make a good commercial decision in the context of numerous information influences. As a result, sellers and customers are increasingly dependent on automated product/service processing, and proxy technology emerged. intelligent proxy is a software entity that represents a user or a program with a certain degree of autonomy to perform certain operations. It operates according to user requirements. intelligence represents the proxy's ability to accept users' purposes and execute tasks. usually, the proxy is signed according to the requirements of the agent object and the currency transaction proxy may be divided into different individuals or organizations with conflicting interests. Therefore, the proxy is not open to the public. it may be impossible for agents to assure all participants of the best results, but the system should at least be as good as the unused agents.

The proxy mechanism has the following advantages:

1) Efficiency: the fundamental starting point of the proxy mechanism is to maximize the benefits of the objects to be proxies, representing the process of related transactions, in order to achieve higher efficiency and better results, as long as the mechanism itself is not worse proxy, proxy can reflect the efficiency.

2) individual rationality: The Role of the agent of the participating mechanism should be at least less than that of the agent of the non-participating mechanism.

3) incentive effect: the mechanism must provide incentives so that the preset actions of agents are optimized.

4) Stability: The proxy should not change its policy due to other proxy behaviors. Ideally. There will be a master policy that is best for every proxy, regardless of the behavior of other proxies.

5) symmetry This mechanism requires no particular proxy. Proxy with similar behavior should achieve the same effect

Intelligent Agents have changed the way businesses are executed, helping to automate a series of activities, saving a lot of time and reducing transaction costs. E-commerce has become more user-friendly, intelligent, and humanized. these features help optimize the entire purchase process. as an e-commerce intermediary, proxy is essentially an intermediate role that enables them to better meet customer information filtering, retrieval, personalized evaluation, and complex cooperation. the most important thing is that they need to communicate with the demand identification, product and seller brokers, and the buy model.

4 case studies

Cisco Systems has benefited from its use of Internet and business intelligence to build new relationships with customers, making Cs7 a successful e-commerce pioneer.

In addition to technical support, Cisco WEB sites provide a wide range of product configuration and order services, including order status tracking systems. all these services are provided by different agents. For example, the Time Agent helps the customer to spend a small amount of time selecting Cisco products. The invoice Agent allows the customer to browse invoices online. The order Agent allows the customer to access service order information; the contract agent provides the customer's contract information; the upgrade Agent allows requests for software and hardware upgrades and documents; the notification Agent allows you to set conditions for receiving email notifications; the configuration proxy allows the customer to configure and process orders by step. The configuration proxy should check the compatibility frequently, thus reducing the chance of errors. Once the configuration process is completed, the product information is directly sent to the production workshop, the customer does not have to re-enter the specified information. The status proxy allows tracking of completed orders. As long as the incoming order number is used, the customer can accurately know the product's operating time, processing status, and delivery method.

In short, Cisco online, which attracts business intelligence, is automated, and has achieved outstanding performance thanks to the on-screen customer support and the online ordering system that saves Cisco a lot.

5 conclusion

The fundamental motivation of business intelligence is to "make better decisions faster ". As a tool or mechanism to improve the efficiency of business operations and improve the effectiveness of business operations, business intelligence has adapted to the needs of e-commerce development and has been widely used. bringing business intelligence tools into the eyes of e-commerce can reveal potential business opportunities behind data. Using Data Mining, neural networks, intelligent agents, and other technologies can help you make correct and rapid business decisions, and make related activities easier and more automated.