CRM系统：基于OLAM技术的分析型CRM设计与开发_英文_
Design and development of ana lytica l CRM
ba sed on OLAM technology
Zu Q iaohong L iW enfeng
(School of L ogis tics Engineering,W uhan U niversity of Technology,W uhan 430063, China)
Abstract: Some key technologies of actualizing customer relationship management (CRM ) systems are
researched. The customer analysis m ining p rototype systems on the basis of on2line analyticalm ining (OLAM )
is designed. After transaction analysis, the data warehouse of CRM is constructed. The CLV /CL /CC customer
division model based on customer lifetime value, customer loyalty and customer credit is emphatically
researched. Three parameters of customer division—customer lifetime value, customer loyalty and customer
credit—are calculated by corresponding algorithms, which can realize customer divisions effectively and
imp rove the accuracy of distinguishing among customers. The data of p roduct sales are analyzed by the sequence
association rules algorithm, the potential rules of the p roducts relevance are discovered, which can p rovide
evidence for supporting decisions such as p romotion strategies. The transaction data such as p roduct sales
volumes and order lists are analyzed on2line through multi2dimensional and multi2level up2drills, down2drills,
and horizontal / longitudinal sections. The customer p roperty factors are analyzed as well. The theory and p ractice
of OLAM and its visualization are further exp lored.
Key words: on2line analytical m ining (OLAM ); analytical customer relation management (CRM ); data ware2
house
Rece ived 2007209210.
Foundation item: The National Key Technology R&D Program of China during the 11th Five2Year Plan Period (No. 2006BAH02A06).
Biography: Zu Q iaohong (1970—) , female, associate p rofessor, zuqiaohong@whut. edu. cn.
O n2line analyticalm ining (OLAM ) , an important kind of know ledge discovery method w ith the on2line analyt2
ical p rocessing (OLA P) fram ew ork, combines w ith the data m ining algorithm. It can obtain useful information and
know ledge from massive, incomp lete, noisy, fuzzy and random data, w hich can be app lied to macroscop ic auxiliary
decision2making and enterp rise development[1 - 2 ].
D ata m ining usually depends on w ell2organized and w ell2p rep rocessed data sources, and the quality of data
sources can directly influence the m ining results. Prep rocessing data sources at the early stages are very important
for data m ining. The OLA P tools of a data w arehouse can conduct various kinds of multi2dimensional data analyses
and integrate w ith data m ining so as to realize flexible and interactive m ining in a number of abstract layers[3 - 4 ].
The data m ining system combined w ith OLA P tools is so flexible and interactive that it can imp rove quality and the
efficiency w hile m ining. It is uniform ly designed in data m ining, data w arehousing and on2line analysis, w hich not
only leads to a great developm ent of the data w arehouse system, but also can be w idely used in p ractice.
1 OLAM M echanism
The key p roblems of realizing the OLAM m echanism are quick response and effective realization in large data2
bases or w arehouses. A n follow ing factors should be taken into account:
1)Q uick response and high performance m ining
G radually refined quality of the OLAM data m ining method: A n interesting model is identified using the fast
m ining algorithm in large am ounts of data, and a higher cost but m ore accurate algorithm is analyzed in detail.
2)M ining method based on data cube
The m ining method based on the data cube w ith high performance is the core of the OLAM mechanism. Effec2
tive computation of a multi2character data cube as w ell as the non2traditionalmetric and comp licated dimension data
cube is important in data m ining[4 ].
3) Interaction betw een m ultip le functions of data m ining
OLA Pps advantages lie not only in the choice of a series of data m ining functions, but also in the interaction
betw een data m ining and OLA P[5 ].
4)V isualization tools
In order to effectively disp lay the OLA P results and interactw ith m ining, it is very important to app ly multip le
know ledge and data visualization tools, such as charts, curves, decision trees and rules maps, cube view s, etc[6 ].
5) Expansibility
The OLAM system, w ith the user and know ledge visualization packages at the top, comm unicates the data cube
at the bottom. The OLAM of the comp licated data including structured data, sem i2structured data and unstructured
data is also an important research direction.
2 A na lytical CRM B ased on OLAM Technology
In the analytical CRM system, the core technology is on2line analysis and data m ining. Through data analy2
zing, some useful rules can be found in m uch of the data through a variety of m odels and algorithms, w hich can
support decision2making. B y clustering and classification technology, custom ers can be grouped. A ssociation analy2
sis theory is mainly for analyzing customer associations, for examp le, w hether a high2value customer base is the
high expense rate customer base or not. Pattern recognition technology is used to identify the particular behaviors of
customers. Forecasting technology is used to find customer future behavior rules. For instance, valuable customer
loyalty degrees can be p redicted by the method of fuzzy neural netw ork.
In analytical CRM, the data m ining techniques can be used in the follow ing app lications:
1) Customer value analysis. Through analyzing the customerspcontribution to the enterp rise business, the cus2
tomer value can be calculated, so different services can be p rovided to customers of different groups[7 ].
2) Customer development analysis. Through analyzing the customersp contribution to the p roduct business, a
relatively accurate reference can be p rovided to a target customer base for p roduct marketing.
The customers can be grouped into the easily loss of valuable custom er base, valuable and stable customer
base, low 2value and unstable customer base, low 2value customer base and stable customer base. A company can
adop t different services, market and p rice strategies to stabilize valuable custom ers, convert low 2value customers,
and elim inate no value customers[8 - 9 ].
The analytical CRM system structure is designed based on OLAM technology (see Fig. 2); then, system func2
tion modules are designed.
强力推荐：
天柏客户关系管理系统
天柏客户关系管理系统（CRM）是一款集专业性、实用性、易用性为一体的纯B/S架构的CRM系统，它基于以客户为中心的协同管理思想和营销理念，围绕客户生命周期的整个过程，针对不同价值的客户实施以客户满意为目标的营销策略，通过企业级协同，有效的“发现、保持和留住客户”，从而达到留住客户、提高销售，实现企业利润最大化的目的。通过对客户进行7P的深入分析，即客户概况分析(Profiling)、客户忠诚度分析(Persistency)、客户利润分析(Profitability)、客户性能分析(Performance)、客户未来分析(Prospecting)、客户产品分析(Product)、客户促销分析(Promotion)以及改善与管理企业销售、营销、客户服务和支持等与客户关系有关的业务流程并提高各个环节的自动化程度，从而帮助企业达到缩短销售周期、降低销售成本、扩大销售量、增加收入与盈利、抢占更多市场份额、寻求新的市场机会和销售渠道，最终从根本上提升企业的核心竞争力，使得企业在当前激烈的竞争环境中立于不败之地。
关键词：CRM,CRM系统,CRM软件,客户关系管理,客户管理软件,客户管理系统,客户关系管理软件,客户关系管理系统