3. An Integrated KM Model For Managing Product and Customer Knowledge
3.4 Evaluation Criteria In A Multiple-Assessment Scheme for Data-Mining
In the integrated KM model a multiple-assessment scheme is proposed to deal with the issue of justification. Three evaluation criteria of the scheme are shown in Table 5.
Table 5. The three evaluation criteria in the multiple-assessment scheme
Can data mining technique successfully extract customer knowledge to facilitate NPD?
2.
Which one is the most appropriate, if multiple data mining methods are qualified for clustering customers into segments?
3.
Does the customer accept the web-based survey approach and render a response sufficient for data mining?
1.
A survey response sample size of 1,000 is assumed to be the sufficient threshold for criterion No.1. As for criterion No. 2, it will deal with (1) how to justify K-means, SOM network, and FuzzyART network as qualified data mining methods for clustering operation in market segmentation. And for criterion No.3, we have to answer: (2) in case that all three methods are qualified for the clustering operation, how to decide which one is the most appropriate?
The first problem that arises is: how to determine the optimal number of clusters or the ‘natural’ number of clusters. Defining a criterion function that measures the clustering quality of any clustering solution, and finding the cluster solution that presents the extreme value of the criterion function is one approach to solve the problem (Duda 2001). In other words, clustering operations to come up with solution for different numbers of clusters is performed repeatedly in order to compare the value of criterion function for each clustering operation. If there is a ‘larger gap’ among the values of the criterion function for sequential clustering solutions, then it suggests the existence of a ‘natural’ number of clusters. Among many criterion functions for cluster analysis, this study chooses R-squared (RS), the ratio of SSb (between-clusters variation) to SSt (total variation), as the criterion function, and SSt
= SSb + SSw, while SSw is the within-cluster variation. Plotting the R-squared values as a vertical axis against the number of clusters as a horizontal axis, an ‘elbow’ point in the plot indicates the presence of the best clustering solution or the optimal number of clusters (Sharma 1996). The above-mentioned procedure is summarized in the Table 6.
Table 6. The procedure to find the optimal number of clusters
This study choose of (between-clusters variation) to (total variation), as the criterion function, and SSt = SSb+ SSw, while SSwis the within-cluster variation.
3
If there is a “ n the of the criterion functionamong then this suggests a “ umber of clusters.
2
Plotting the R-squared values as a vertical axis against the number of clusters as a horizontal axis, an “ point in the plot indicates the best
clustering solution or (Sharma 1996).
4
Defining a criterion functionthat measures the of any clustering solution.
This study choose of (between-clusters variation) to (total variation), as the criterion function, and SSt = SSb+ SSw, while SSwis the within-cluster variation.
3
If there is a “ n the of the criterion functionamong then this suggests a “ umber of clusters.
2
Plotting the R-squared values as a vertical axis against the number of clusters as a horizontal axis, an “ point in the plot indicates the best
clustering solution or (Sharma 1996).
4
Defining a criterion functionthat measures the of any clustering solution.
The second problem will be resolved by adopting the existing market research paradigm: using the most commonly used Cronbach’s alpha coefficient to represent the reliability of construct in a rating scale survey instrument (Cronbach 1951, Peterson 1994).
Cronbach’s alpha is also an estimator of internal consistency for a survey instrument, which is approximately defined by the formula:
r
where K is the number of items in the survey instrument, r is the average inter-items
coefficient of correlation among the items. And is the variance of all values in item i, and is the variance of the total item values for each customer response.
ó2i
ó2s
If the inter-items correlations are high, it can be inferred that the items are measuring the same underlying construct. It also means that the internal consistency or the reliability of the survey instrument is sufficiently reliable for business practical utility.
After market segments is formed by segmentation task, all the survey response will be re-allocated to each segment, and calculate Cronbach’s alpha of all the instruments belong to each segment to make sure whether the data mining method is practically useful in real world business environment. The recommendation for the minimally acceptable reliability level expressed by Cronbach’s alpha is proposed by researcher as in the Table 7.
Table 7. Recommended reliability levels for rating scale survey instrument
Researcher Situation Recommended level
Murphy and Davidshifer (1998)
Unacceptable level Low level
Moderate to high level High Level
Below 0.6 0.7 0.8-0.9 0.9 and above
To summarize, the whole procedure for the multiple-assessment scheme is shown in Figure 9. An empirical study described in the Chapter 4 will furnish the field-collected data set to go through the multiple-assessment scheme. Data mining results assessed by the three evaluation criteria will confirm whether or not it is robust to be used in the integrated KM model.
CHAPTER 4
IMPLEMENTATION OF THE INTEGRATED KM MODEL
4.1 BACKGROUND ABOUT THE CASE STUDY
To meet future demands on Mobile-commerce, an innovative NPD project was initiated. The objective of this project is to explore the opportunity for Taiwan’s hi-tech industry especially the computer and wireless communication hardware manufacturers to penetrate into a new market -- the in-vehicle personal computer market. The vision of this NPD project is to serve perspective users beginning from Taiwan and extending to the greater market coverage in the world by taking advantage of economies of scale and economies of scope. Aiming at project success, the integrated KM model with an empirical study has been implemented in this innovative NPD project, in view of linking customer knowledge with technological product knowledge and dealing with the complexity of product sophistication/market acceptance issues.
4.1.1 Wireless Internet as The Technological Innovation to Support Telematics and Mobile-Commerce
In this era of the digital economy, companies have benefited from the capabilities of the Internet to operate sales and communication channels in order to reach customers in larger geographical regions. This new approach is titled ‘Electronic-commerce (E-commerce)’ -- selling and buying products/services over the web. In the E-commerce, the connectivity within the fixed user locations in the wired communication infrastructure can be further enhanced by the innovative ‘ubiquitous networking’ wireless communication technology, to construct the ‘Mobile-business business)’ or ‘Mobile-commerce
(M-commerce)’ environment under the definition of “M-business = Internet + wireless + E-business” as given by Kalakota (Kalakota 2002).
The computer-based M-commerce platform in a vehicle, namely the automotive Telematics platform, (Telematics is the combination of “Telecommunication” and
“informatics”) is regarded as an innovative technology combining wireless communication, an in-vehicle information system, and an in-vehicle multimedia computing system. A Telematics-enabled vehicle is capable of offering customers a variety of benefit and value in the form of new services, namely, the enhancement of safety, security, travel information services, convenience, and entertainment, so as to develop new business models and new markets (Figure. 10). Therefore, Telematics is qualified as a radical innovation enabling technology if taking into account the benefits it renders, the sophistication of its engineering efforts, and the complexity of application in wireless network connectivity.
However, a recent report demonstrated that customers within different territories presented heterogeneous needs patterns: the customers in the U.S. have much stronger focus on safety-relevant services, while those in Europe prefer navigation and traffic information services, and customers in Japan accept more entertainment-based Telematics packages (Shahmanesh 2003). This implies that in order to develop a Telematics platform for market success in a particular region, the sophistication of technologies should not be the only issue to deal with, but consideration on the customer side also matters. Therefore, building a M-commerce Telematics product demands innovative technology as well as customer knowledge. It is also worthwhile to note that M-commerce is different from the endeavor of attracting the mass market in E-commerce, scholars have suggested that in M-commerce the emphasis should be put on separate service packages for different specific groups of customers (Kotler 2003).
D ig it a l C o n t ent s Pro v id er
-M-commerce model com bining wireless communication, multimedia, and IT
T el em at ic s
4.1.2 The Backbone of M-Commerce – The ‘Ubiquitous networking’ Technology To realize M-commerce applications, technology innovation is definitely a must in addition to the elimination of impediments over current E-commerce including data security, legal affairs, commercial settings, cost-benefits, as well as the context for user-friendly application.
Once equipped with wireless Internet access capability, a cellular phone set, a personal digital assistant (PDA) and a portable notebook computer become all qualified devices for M-commerce. They allow users to ubiquitously access the Internet at any time and any way through various wireless networks operating within different distance coverage via a Bluetooth wireless network, the wireless local area network (IEEE 802.11b), the wide area network like GSM (Global System for Mobile Communication), GPRS (General Packet Radio Service), and W-CDMA (Wide Band Code Division Multiple Access) network. Unfortunately, compared to a wired Internet system, wireless devices
feature two kinds of drawbacks: the first is the low-bandwidth and high-latency time, the second is usage constraints such as short battery storage capacity, small monitor screen size, and erratic communication reliability under different operating environments, let alone the costly wireless connecting expense.
The battery and screen issue does not happen in a computer-based M-commerce platform in a vehicle, namely, an automotive Telematics in-vehicle platform. (Of course, this is at the expense of sacrificing the mobility of a cellular phone). It is engaged in the perspective of M-commerce by getting the car industry into the Internet era, and reciprocally bringing the Internet into the car by wireless technology (Sommerlatte 2001).
4.1.3 Building An M-Commerce Platform Demands Innovative Technology and Innovative Knowledge
The M-commerce model supported by a Telematics in-vehicle platform demands the application of a variety of high technology. When integrated as a whole the task becomes so sophisticated that many efforts need to be made.
Quoted from Fuchs (2002), the major enabling technologies of Telematics are:
(1) Positioning and location technologies: GPS (global positioning system), digital map.
(2) Telematics service delivery technologies:
Wireless short-range communication system: Bluetooth, IEEE 802.11x networking family.
Wide area cellular communication system: GSM, GPRS, CDMA, W-CDMA.
Communications via satellite.
Data broadcast: FM subscriber, digital audio broadcast, digital video broadcast.
(3) Networking and protocols Telematics protocols.
Internet protocols.
WAP (wireless application protocol).
(4) Vehicle communications
Serial vehicle bus systems: CAN (controller are network), LIN (local interconnect network).
Multimedia/plug-and-play bus system: MOST (media oriented systems transport) and AMI-C/IDB (automotive multimedia interface collaboration/ ITS data bus) Audio and speech processing
(5) Distributed computing
4.2 RESEARCH METHODOLOGY FOR MODEL IMPLEMENTATION
The implementation of the integrated KM model is supported by an empirical study in an innovative Telematics NPD project. It begins with managing product knowledge to facilitate the identification of product benefits, based on which to develop a web-based market research for customer knowledge management. Research methodology involves data mining by statistical and neural network approaches that are employed to extract knowledge of customers in marketplace, as well as a multiple-assessment scheme that is used to justify research outcomes in this dissertation.
4.2.1 Product Knowledge Management for Telematics NPD Project
The sophistication of innovative technology involved in Telematics NPD project has been revealed in section 4.1.3, the confidential nature of related product knowledge prevent
itself from being further disclosed. Therefore, in this chapter we play down the detailed discussion of product knowledge management and rather put emphasis on the aspect of customer knowledge management.
4.2.1.1 Organizational Learning
It is certain that no business entity can afford the possession of all the above-mentioned knowledge and technologies to alone conduct an automotive vehicle-installed Telematics NPD project. Given that all the technologies become available, there is still no opportunity to secure market viability if perspective customers turn down the offerings by judging them as premature. It is obvious that aside from the working knowledge on innovative technologies, the NPD team should create measures of how to exploit knowledge hoarded in value chain partners and measures of how to penetrate the market.
Organizational learning is required for knowledge sharing and knowledge generation among team partners at many interfaces. ‘Hierarchical interface learning’ scheme is employed in this project to tackle the demand that learning must occurs at the interface between marketing and product design, the interface among value chain stakeholders, the interface among heterogeneous technologies, and so on. From product technology viewpoint, ‘hierarchical interface learning’ may start from areas such as the interface between the wireless communication modules and computer main board; interface among vehicle electronic harness, control console, operating software; and interface between human-machine interaction and driving safety.
4.2.1.2 Value Chain Collaboration
The players, technologies, and activities involved in the framework of M-commerce consist of two domains with three processes in each domain. In the contents domain, the processes are: (1) contents formats: text, hyper-text and graphic, audio and video, (2) contents presentation: media editing and presentation, value-added website set-up, (3) contents for commercialization: contents and services providing, mobile wireless portals for commercial practice. On the other hand, in the infrastructure domain, the processes are (1) mobile wireless transmission technologies and telecommunication companies, (2) mobile services/delivery support: Internet connection, server platform, payment system, and security measures, (3) mobile interface/application: application based on embedded operating system device hardware, firmware, and software (Pierre 2001). The device to activate the transaction is through portable computer, PDA, or even a mobile phone.
In case of in-vehicle Telematics application, aside from the above-mentioned M-commerce activities, additional customer-attractive functions and vehicle-related services become indispensable to satisfy car drivers with specific demands. Therefore, the first thing is to organize a cross-industry collaboration NPD team to develop a car-installed Telematics platform, and then the next step will be the establishment of conciliation among value chain stakeholders to find out how to do business.
The ‘collaborative engineering’ approach enables the distribution of within-organization and between-within-organization tacit knowledge among NPD team partners to facilitate the generation of new knowledge in how to construct the building blocks of this innovative product, and the resolution out of consensus will be codified for application and for reuse in the future. An example of generated knowledge is the application difference between personal computer operating system such as Window 2000 and embedded operating system such as WinCE.Net, to affect the wireless Internet connectivity capability.
4.2.1.3 Technology Transfer
As a consequence of collaborative engineering, the practice of modular design as an existing knowledge and technology in the computer industry has been transferred. It alerts the NPD team to conjecture whether a full-featured new Telematics product is likely to be judged as premature if customers perceive it as too costly, too sophisticated, and too inconvenient. Therefore, the NPD team makes a proposal to apply the knowledge of product ‘re-configurability’ that can be embedded in the ‘modular design’.
To implement the ‘modular design’, each party in the NPD collaboration team transfers only the technology that is required at the interface of different domain knowledge to let NPD product comes into being. The ‘hierarchical interface technology transfer’ scheme provides the appropriate solution for technology transfer at interfaces of different product module or platform. The codified product knowledge is used to make up the ‘value chain knowledge/technology architecture’ as the blueprint to guide NPD team in accomplishing all the necessary knowledge/technology transfer.
4.2.1.4 Product Feature Realization–The Outcome of Product Knowledge Application The NPD collaboration team adopted herein a multi-layered architecture to accommodate all the features of Telematics platform. Presented here only for demonstration is an oversimplified two-layered architecture example. At the system level, a 2 x 1-DIN (Deutsch Industry Norm) size box or 2-DIN size box is designated as two alternatives for the first layer configuration. In the second layer subsystem, there are four main building blocks: wireless communication module, multimedia module, GPS navigation module and HMI: human-machine-interface including a 7-inch TFT (thin film
transistor) display and control panel either with or without voice-command hand free phone capability.
The presumed full-scale system is the whole of wireless, multimedia, GPS, and HMI modules. In case of control panel without voice-command hand free phone capability, product variants become available by a combination of: (wireless + HMI), (multimedia + HMI), (GPS + HMI), (wireless + multimedia + HMI), (wireless + GPS + HMI), (multimedia + GPS + HMI) and (wireless + multimedia + GPS + HMI). Each variant has specific features and cost structure to meet the demand of particular customer group, and each also conserves the possibility of re-configurability when technology evolves in the future. At this stage, knowledge realization comes true.
4.2.2 Customer Knowledge Management for Telematic NPD Project
This part is the main focus of this dissertation, backed up by a web-based survey.
4.2.2.1 Product Benefits Identification
At first, the NPD project team generates and delivers product knowledge ‘for’
customers. Customer knowledge generation starts from transforming by NPD team the Telematics product’s features into ‘benefits perceived by customers’, paving the way to understand customers’ response toward benefits that those features bring out. The typical benefits of the Telematics platform for the vehicle drivers are: safety, security, information, convenience, entertainment and M-commerce application. Table 8 is a summarized but not exclusive list describing features that Telematics platform has, benefits it offered and the technologies involved. At this stage, a bi-directional communication channel like E-mail or company’s website is appropriately used to reach customers.
Table 8. Telematics platform’s features, benefits offered and technology involved Features Technology Used Benefits Offered
Crash notification GSM, GPRS Safety
Emergency services GSM, GPRS Safety
Roadside assistance GSM, GPRS Safety Hand-free mobile
tracking GPRS or GPS Security
Vehicle location change alert
GPRS or GPS Security
Vehicle tow service GPRS or GPS Security Automatic navigation Tourist information GSM, GPRS, Bluetooth,
Wi-Fi (IEEE 802.11b)
Information, Convenience Traffic condition report GPRS, Bluetooth,
Radio broadcasting Information,
In this study, sampling plan for web-based survey is based on the following assumption:
(1) Because Telematics features are complementary and supporting devices for a vehicle, only vechicle drivers to be the potential customers of the Telematics platform.
(2) Only customers those are used to IT application such as E-mail, Internet surfing and on-line conversation, to be the perspective customers of the sophistiscated Telematics platform.
(3) Only customers with enthusiasm about high technology stuff are interested to answer the web-based survey, their response toward Telematics platform will be more optimistic than the average customers.
Under the assumption, the sampling plan is prepared by a random selection from the current customer database to form a target roster, by which to send out an e-mail soliciting the recipient to answer a self-administered questionnaire posted on the website www.question.sine.com.tw. On the front page of that website appears a thank-you letter, followed by a detailed instruction describing the procedure how to answer the questions.
The first part of the questionnaire is pertinent to customers’ demographic and personal information such as age, gender, profession, education background, income, marital status, and number of children. The second part of the questionnaire in the survey instrument consists of 29 items of Telematics features that deliver certain benefits to customers. All 29
The first part of the questionnaire is pertinent to customers’ demographic and personal information such as age, gender, profession, education background, income, marital status, and number of children. The second part of the questionnaire in the survey instrument consists of 29 items of Telematics features that deliver certain benefits to customers. All 29