Ph.D thesis "The Viability of Multimedia Retrieval Systems for Marketing and Sales" by prof dr M. J. Hoogeveen  

3.9. The Virtual Market (VM)

3.9.1. Introduction

A free and public Virtual Market (VM) is realisable using MT. The VM is a global and public electronic market place, that has a transparent user interface, where demand can meet supply freely. An attempt is made, within the EURESCOM IMS1 (Integrated Multimedia Services at 1 Mbit/s) project, to specify a generic MM service (Simon, 1993; Hoogeveen, Lecoq & Balestri, 1993); this can be seen as an MM successor to Videotex (Hoogeveen & Andersson, 1993), and can be compared to a VM service. The VM service becomes a real VM when demand and supply meet each other freely, the VM infrastructure offers public access, and a low threshold for accessing the market.

The VM exists already, but is fragmented and in its infancy. To give examples: there are countless numbers of bulletin boards offering information products like games and pictures, electronic book shops, POSs offering a VR user interface of a shopping centre, etc. Experiments by AT&T and Silicon Graphics in Orlando with TV-I will provide information about the technical feasibility and viability of setting up a VM service.

What types of M&S applications can be found in the VM? In the history of Videotex most applications fall into the categories banking and insurance, retailing and distribution, electronic publishing and travel agencies. (Online, 1984). In an MM successor new types of applications will probably be added that include video and/or audio data streams, like video on demand.

In principle there is no limit to the number and types of applications that may be offered in the VM. There will naturally be a kind of self-selection due to commercial viability of an application. An overview of research in MM telecommunication gives a good idea of the types of applications to expect in the VM. Research is performed on MM telecommunication systems for education (Gullickson-Morfitt et al., 1992; Mühlhäuser, 1991), MM teleconferencing (Crowfort et al., 1991), computer supported co-operative work based on broadband (Kindt et al., 1991; Pehrson et al., 1992), electronic publishing (Consulting Trust, 1993), medical imaging (Vöge, 1992), video on demand (Venkat Rangan et al., 1992), financial information services (Arbuthnot & Khalil, 1993) and in many other areas of application.

After the introduction of the first generation of VM services, the next generation VM services will gradually evolve and improve. The continuous confrontation between demand and supply, and service competition between information providers will lead to the deselection of information services that have little value added and a low quality of service. The struggle for survival will sift the viable from non-viable information services. The creativity of human kind may result in yet unforeseen types of information services.

A gradual increase in the number and quality of MR products and MR services can be expected. These products and services will become more and more self-selling, making the sales staff of suppliers superfluous, to a certain extent. The VM will grow from national to continental to global, and will stimulate further the globalisation of the national and continental economies. It is possible that the economic value of the VM will grow to such an extent that it will overshadow the economic value of traditional non-electronic marketplaces.

The following descriptions of the M&S business objectives and information processing tasks, MRS functions and MRS implementation aspects are based on participation in the IMS1 project, and make use of knowledge of state of the art technology.

3.9.2. Business objectives

The business objectives must be seen from the points of view of the different actors in the VM place:

• the customer or purchase department;

• the information provider (IP), especially its sales department;

• the service provider (SP);

• the network provider (NP);

• the financial parties (FP).
 

The last three actors, the SP, NP and FP, are in fact facilitators. They are considered here explicitly because they are continuously and actively involved in keeping the VM service in the air and offering feedback on use of services and financial transactions as an integral part of the VM service. The business objectives (only in the case of private customers is it often not possible to speak of business objectives) are discussed per type of actor, below. Per actor the introduction bottlenecks, that form a risk for the viability of the VM service and for meeting the business objectives, are mentioned shortly. The facilitators SP, NP or FP can also act as an IP, but this will not be discussed here.

Two categories of customers may purchase the products and services of an IP: consumers or purchase departments. (Possibly the purchase departments of other IPs). There will be both professional and non-professional users.

The main objectives for the customers are:

• reduction of purchase costs by efficient access to IP services or by discounts;

• meet demands not met before by being able to purchase, for example, new types of services like a personal news magazine;

• more choice between suppliers and their information services and products;

• improved infotainment (in case of consumers).

A number of introduction bottlenecks can be seen in relation to the customer agent:

• the availability of an affordable and flexible bandwidth;

• the availability of affordable equipment;

• business value added of information services for customers;

• end user acceptance, which depends on the usability of the VM (applications) and resistance to change.

 

The IPs in a VM are those companies or persons who have invested in a certain application, in a certain information service. An information service consists of the information provided and the functions programmed to handle this information.

The business objectives for IPs are to:

• (depending on the number of VM access points) extend their market penetration;

• increase the sales volume;

• reduce marketing & sales costs by making retailers and M&S personnel superfluous;

• obtain immediate and precise information about customer behaviour;

• obtain direct access to their markets and direct contact with their customers;

• improve quality of service by reducing delivery delays of information products.

Introduction bottlenecks for the IP are related to:

• the initial costs of setting up an information service;

• the difficulty of predicting in advance the success of an information service which depends on actual use by customers;

• lack of experience with an MM service;

• copyright protection of information provided.

SPs consist of the parties who offer the VM service to VM service subscribers. SPs maintain the information servers, the connections to the information servers and may be involved in the production of an application or specific information service.

The business objectives for the SP are to

• set up a cost effective VM service, to be profitable;

• to generate extra revenues for the shareholders, often NPs, some large IPs, existing SPs like Videotex and Minitel, and perhaps FPs.

Introduction bottlenecks are:

• lack of experience with MM service provision, e.g., multi-user aspects in relation to real time use of an application;

• uncertainty about Return On Investment (ROI) for SPs: when is the market ready for a VM service and in what form?

 

NPs play second fiddle in the VM concert. Their role is to provide the network connections between the entities involved, to offer - simply said - bandwidth, and to take care of network accounting.

The business objectives for participating NPs are to:

• generate more network traffic;

• offer new types of network services (Value Added Network Services);

• defend or extend its market share.

FPs, like credit card organisations or banks, need to open their financial infrastructures to the VM service to support real time financial transactions between, most importantly, customers and IPs.

The business objectives for FPs to support VM transactions are to:

• increase their market share by offering VM transaction processing services;

• increase efficiency by further automation of transaction processing.

An important introduction bottlenecks is:

• uncertainty whether there will be an acceptable ROI for setting up VM financial services.
 

A VM is only viable if the most important objective of these different types of actors are met: an acceptable ROI. Only if the VM services are clearly profitable for potential parties can they be persuaded to participate in a VM service consortium and in VM applications.

3.9.3. Information processing tasks

The VM is always part of a real market (see figure 28). In the foregoing paragraphs the roles of the customers, IP, SP, NP and FP are described in general terms. Their specific interactions with the VM are described in this section.

Three basic tasks of customers and suppliers are distinguished: purchase, production and sales (promotion and delivery). Firms can have both the role of a customer and a supplier. All suppliers that are involved in the VM act as an IP, because they provide information about their products and services.

The purchase department or customer browses through the information services, identify themselves (optional), and place orders. In return customers pay for the use of the electronic information services and the delivery of products. Delivered products can be electronic information products like electronic reports or home videos. Non-electronic products and services are delivered physical by the IP. The customer receives invoices and bank statements from the financial information service that is part of the VM service.

A purchase department feeds a production department with the semimanufactures that they have purchased. Non-electronic semimanufactures may be send on physically, electronic semimanufactures may be delivered by an MM mail service, such an MM mail service can be provided by a mail company acting as an IP or by the SP as a standard service. Production departments distribute, in their turn, their manufactures either electronically or physical via their sales department.

The sales department is recognisable as an IP, since it offers the information service in the VM. The IP has the possibility to update its own information service on-line, but it may also farm this out to its production department, a subcontractor or a service provider. The IP monitors the use of its information service, receives orders from customers, sends invoices, and receives payments from customers via financial parties.

The tasks of the SP, that has set up the VM, are to monitor use of VM services by all the other actors, to send invoices to these actors for the use of the VM services, and to update continuously the information services. Updating the service can mean maintenance, adding new information services or new functionality to the VM service, and removal of outdated information services or obsolete functionality. The SP also offers a production environment as a paid facility for IPs. The information services are produced in the production environment. Therefore content data needs to be created, gathered, edited and prepared digitally. Next, the information services are authored, tested and transposed to the service environment.

NPs view a VM service as a specific VANS (Value Added Network Service). The main tasks of a NP consists of offering variable bandwidth connections between access points for users, hosts, and the service management centres. Further, the NP monitors network use constantly to optimise data traffic, and to account and bill for use of the network. Accounting and billing may be integrated with the accounting and billing for the SPs and/or IPs. Payments of VM users will be based partly on a fixed connection price and the price of network use. (The price for the complete service will also depend on the prices of information service use and products provided).

An FP receives all orders for financial transactions: from customers that have purchased a service, from IPs that order automatic payment or pay the SP or NP, from the SP and NP ordering automatic payment and settlements among themselves. The FPs perform financial transactions and give feedback in the form of bank statements, possibly via their own banking information services.   Figure 28. Roles and processes in relation to the Virtual Market (VM).

3.9.4. System functionality

A description of the functions of the VM service (Jones, 1993; Hoogeveen & Van den Eijnden, 1994) is given. First, the general VM service functionality (see figure 29) is discussed, then the VMSE functionality (see figure 30).

  Figure 29. Functional elements of the VM.

As can be seen in figure 29 several types of system environments are present in the VM; the term 'agent' is used to indicate that these environments can, to a certain degree, act independently and be intelligent. The types of environments are the customer agent, IP agent, service provider agent, Information Service Production Environment, network management agent, financial agent, and the most central environment, the VM Service Environment (VMSE).

• The customer agent offers access to the information services of the VM. The customer agents supports personal identification, order entry, navigating through the VMSE, using information services and payment for services used.

• The IP agent consists of functionality to monitor the use of its information services, to make small changes to its information services (e.g., repricing products), to accept orders from customers and payments via FPs. The IP agent may be connected to internal order administration and invoicing ISs, and logistic ISs, so that order handling is mostly automated.

• The SP is responsible for one or more service environments (VM nodes), support of the customer environments, and may offer an Information Service Production Environment (ISPE) as a paid facility for IPs. There may be many distributed VM service environments managed by one or more service providers. The SP agent supports the SP in monitoring use of services, tuning the VM service environments, and invoicing related to use of services.

 • In the ISPE the information services are produced for the VM. It is important to notice that we still speak of information services whilst most of these will be levelled up to MR services. For the production of information services content data need to be created, gathered, edited and prepared digitally. The information services also need to be authored, tested and transposed to the VM service environments. Three types of principals for the ISPE are presented in our model (see figure 29). The IP sales department has the possibility to update his own information service on-line. The IP also may farm this out to a production company or department with an electronic publishing facility. The IP can enter into the agreement with the SP, that the SP is responsible for production and maintenance of the information service.

 • The NP agent monitors network use, optimises data traffic, detects queuing problems, reroutes data traffic, etc. Invoices for the use of the network are sent to the SP, who on-charge network costs to its service subscribers, or are sent to the financial agent.

• The Financial agent supports financial transaction handling. Transactions are received from the VMSE, are processed by the banking systems of the financial parties and are returned in the form of electronic bank statements and payments. Bank statements and payments are also registered in the VMSE.

Typical functions in these distributed VMSEs include:

• VM directory access: the VMSE interface offers the ability to present directory information about the information services provided. Directory information is extracted from the VMSE distributed database system.

• Information services access: a customer agent is offered access to information services selected from the directory. A dynamic interaction between the information service stored in the VMSE distributed database system and the customer agent is supported when an information service is used.

• VMSE monitoring: IP agents and SP agents particularly are interested in making use of VMSE monitoring functionality to present service use information. This information includes information about the number of transactions per service or subscriber, the time used per service or customer, and user interactions to analyse customer behaviour.

• Financial transaction manager: the financial transaction manager keeps track continuously of transactions made during customer sessions. The parties involved are informed. The financial transaction manager further processes and sends on invoices.

• VMSE distributed database system: the VMSE distributed database system is the core functionality, that manages the distributed databases containing information services, use of services information, and directory information. Information service updates come from ISPEs and SP agents. The FP updates its financial services with new bank statements about payments.  

Figure 30. Functional elements of the VMSE.

3.9.5. Implementation aspects

The MRS implementation aspects for the VM service are discussed here. First, general MRS implementation aspects are discussed per environment (see figure 29), then MDBMS implementation aspects.

There are two main ways to realise a VM: one, gradually update existing information services, like Videotex, or two, set up a completely new, and for its tasks optimised, environment, that is downward compatible with older services, and offers for that purpose Videotex/Viditel emulation. The choice made is a political and economical decision for the investors involved, and this is not elaborated further in this thesis.

The customer agent offers access to the information services of the VM. The hardware platform for the customer agent consists of a terminal with graphic display (PC or TV based), a flexible public network connection, local storage capacity (many GBs hard disk or CD-MO), multi-card reader (for cards like credit cards, bank cards, telephone cards, etc.), a decoder for MM data streams, audio and video boards, loudspeakers, printer, a video camera (an 'eye' on the screen), and a built in microphone. Depending on customer requirements a tailor-made hardware platform can be devised.

The software component of the customer agent includes:

• the terminal operating system, e.g., DOS and MS-Windows for a PC or RTOS for CD-i, can be used with its related functions, including software drivers for the video overlay board, JPEG and MPEG decoder, multi-card reader and network card, and drivers for MM presentation;

• the VM shell, a preferably intelligent user-interface, that is activated by the operating system shell, and that contains general VM functions for personal adjustments, accounting and billing, accessing the VM service directories, obtaining general help, copying information to a personal directory or printing certain information, controlling the remote connection, changing the set-up, for example the language, and quitting the VM shell;

• local applications like a simple MM editor.

Serious problems that need to be solved at the customer agent terminal are clock synchronisation, out-of-sequence data packets, packets to packets jitter, and lost data packets (Richard et al., 1993).

The IP agent and SP agent are roughly similar to the customer agent, however, the SP agent in particular makes use of a high end terminal as the management of systems requires heavy processing power.

An application development environment needs to be composed for the production: the ISPE. The ISPE includes, in comparison with a customer agent:

• heavier computer platforms for processing MM data;

• access to specific information services in the VM for the retrieval of electronic semimanufactures;

• access to storage media using a D1 and/or VCR recorder, a DAT and/or DCC recorder, a CD-ROM/XA and/or CD-i player, a high resolution scanner, back-up tape streamers;

• input media like a caption camera for stills;

• digital processing including A/D converters, audio/video/still/graphics editors, audio/video special effects generator, MPEG1 encoder, JPEG encoder;

• MM authoring tools for the production of an MM information service;

• a customer agent simulator to test how an information service behaves in the customer environment and other agent simulators in so far as different from the customer agent.

The production of video and audio will probably be contracted out to audio studios and video production companies such as Valkieser in the Netherlands.

As discussed in the previous section it is the task of the NPs to offer the necessary bandwidth for VM sessions. They need to 'create' a virtual VM network (1.544 Mbit/s is enough for real time MPEG1 data stream with video and audio), consisting of many interconnected national networks. They need to offer variable bandwidth connections to the users and service nodes in the virtual VM network. This requires that investments must be made to offer at least 1.544 Mbit/s to the home. Alternative possibilities for this are advanced copper systems like ADSL (Asymmetric Digital Subscriber Line) or HDSL (High-bitrate Digital Subscriber Line) (Hoogeveen & Andersson, 1993), and Passive Optical Networks (Vaalen, 1993). ADSL and HDSL can be used in combination with basic rate ISDN (2B + D) on the same twisted pair. Currently, ISDN-30 meets the MPEG1 bandwidth requirement, although it does not meet the efficiency requirement with regard to variability of bandwidth (variable bit rate). The price of ISDN-30 connections is also about 10 times the pricing of basic rate ISDN. An interesting alternative is formed by ATM in combination with PDH, for which standards have been developed.

The main realisation bottlenecks is that:

• at least 1.544 Mbit/s bandwidth to the subscriber is needed which is far more than that offered by standard and affordable switched telecommunication connections today.

After the introduction of a VM service a growth path towards higher video quality, requiring 4-10 Mbit/s for MPEG2, should be anticipated; if multiple video application sessions need to run in parallel manifolds of 1.544 Mbit/s bandwidth are also required. Again, this means that ATM based Broadband ISDN comes into the picture. The results with ATM based MM services are very encouraging (Armbrüster & Wimmer, 1992). An alternative is formed by the possibility of using CATV networks, although huge investments in introducing switching technology need to be made, however, the heralded introduction of the digital super highway, based on fibre to the home would remove the current bandwidth bottleneck.

The VM service must present itself as a whole to the user, although many VMSEs (VM Service Environments) may exist in different countries, which may make use of many information servers per country. The development of operating systems that support distributed MSs would be very useful (Leslie et al., 1992).

Important bottlenecks to overcome are the, compared to a situation with information servers connected to PSTN, up to 30 times higher demands related to throughput, access times of storage media and processing power in the service environments. This requires large investments and distributed solutions.

A VMSE consists of:

• one or more (clusters of) database servers;

• an extendible storage capacity starting with some hundreds of GBs on, for example, RAID (Redundant Array of Inexpensive Disks) or other storage media;

• hundreds of data communication connections offering at least 1.544 Mbit/s each (e.g., in the form of ISDN30 or ADSL connections);

• the MM databases containing the information services of IPs.

With decreasing hardware and transmission costs, the use of MM services in the VM will be stimulated. Improvement of price/performance ratios for storage media and processing power will make it cost effective to improve service environments so that they can handle still growing, very large, databases.

Let us proceed with the discussion about MRS implementation aspects by discussing MDBMS implementation aspects. In table 4 below an overview is given of four scenarios for the VM and the most likely integrational approach and retrieval facilities related to these four scenarios. The four scenarios are related to time: the hybrid scenario seems cost effective now, the short term scenario as yet, does not seem cost-effective, the middle term and long term scenarios (see next section) are more mature and will probably be cost-effective in the next century if a real data highway becomes available.

A first step towards a VM is made in the hybrid scenario. The main starting point for this scenario is that as much as possible data intensive data traffic is kept local on the terminal of the customer, and only less data intensive data traffic uses the network. Examples of such hybrid MR solutions are a PC with access to a CD-ROM for audio and video and a network connection for structured data, or a Tele-CD-i with audio and video on CD-i and a network connection for structured data. Such a hybrid scenario is used to reduce network costs for the customer, and to reduce queuing problems in the VMSEs. Therefore no real DBMS is used, only some search facilities.

In the short term scenario the complete VM databases are located on the distributed VMSEs. MPEG1 video and audio is send over the network using ADSL or advanced copper systems. The relational model is implemented only for some structured data, not for the documentary, audio and video data, since this would require too much processing power in the VMSEs. Further, no differences with the hybrid scenario can be seen, since these would threaten response times (system performance).

Only in the middle term and long term, are retrieval extensions made, when the price/performance ratios of processing hardware and networks have improved significantly. The most striking extensions for the middle term scenario are the full use of the extended DBMS approach making possible the implementation of the relational model, support of MM document standards like HyTime or HyperODA, more free use of search facilities in information services, and the availability of full text indexing. MM editing is also supported, this is necessary to produce MM information items like MR products.

The long term scenario, probably realistic after about ten years, i.e., after 2004, also includes implementation of the object oriented model, the support of pictorial and sound search, and the support of real time encoding and decoding of MM and hypermedia objects (e.g., by MHEG) for real time exchange of hypermedia objects between systems. The presentation facility is improved so that HDTV quality video and CD-DA quality stereo sound can be played.

Table 4. Overview of four scenarios for a VM service and their likely retrieval approach and facilities.

3.9.6. Discussion

In the foregoing paragraphs an overview is given of a VM service, that includes a variety of MM applications which are able to meet a tremendous variety of customer specific needs.

The main potential advantages for the types of actors involved are made clear. The technical bottlenecks, although often complex, do not seem invincible. Experiments and prototypes have shown that VM services are technically feasible (Hoogeveen & Andersson, 1993). The major bottlenecks are related to organisational and financial issues: how to set up a profitable information service? Which applications offer enough value added to be appealing to sufficient(4³Å@ words, short an׺ìÿ”à> @ÿÿÿÿÿÿT! costs will d׺ìÿ á> @ÿÿÿÿÿÿT! be found to c׺ìÿ¬â> @ÿÿÿÿÿÿT! will be ma׺ìÿ¸ã> @ÿÿÿÿÿÿT! @ÿÿÿÿÿÿT! overall pictur׺ìÿÐå> @ÿÿÿÿÿÿT! ISs are devel׺ìÿÜæ> @ÿÿÿÿÿÿT! Of cours׺ìÿèç> @ÿÿÿÿÿÿT! are grouped׺ìÿôè> @ÿÿÿÿÿÿT! SSSs can be׺ìÿ @ÿÿÿÿÿÿT! by the diverse׺ìÿë> @ÿÿÿÿÿÿT! MPSs, ׺ìÿì> @ÿÿÿÿÿÿT! produ׺ìÿ$í> @ÿÿÿÿÿÿT! few connecti׺ìÿ0î> @ÿÿÿÿÿÿT! a corporate M&׺ìÿ<ï> ï> @ÿÿÿÿÿÿT! common information coding and document structuring standards can be adopted to foster unhampered exchange of M&S data later on. Sharing of production facilities and information sources has synergetic effects.

3.10.2. Indications of the viability and the business objectives and success/risk factors per system

In the introduction of this chapter we asked the question what potential viable MRSs for M&S can we distinguish on the basis of practical examples or as extrapolations of developments in M&S, multimedia systems and retrieval engines?. A number of alternative, more or less complementary, MRSs are presented, extensions on these systems are anticipated, and their mutual relations are discussed in an overall system.

The MRSs seem representative for MR developments within larger M&S firms, although permutations in business functions, system elements and retrieval facilities may lead to innumerable variations on the MRS themes.

Are these systems potential viable? All of the MRSs described are feasible with current technology. Nevertheless, there are still some important technical bottlenecks which need to be coped with:

• the price/performance ratio of MM transmission is a bottleneck that will only gradually disappear;

• the lack of standardisation and sometimes too little market acceptance of standards (de jure or de facto) is a bottleneck for system implementations.
 

A number of the MRSs for M&S discussed seem viable if we take into account that some implementations of these systems are already operational, and are accepted by the management and users involved. This is the case for the IECT photo archive (MCA), the TSA system, the MPS, and MAI systems. To gain more insight into the viability of MRSs for M&S it seems necessary to measure to what degree stated business objectives are met. Only in a very few cases are IT evaluation methods used, and when an IT evaluation method is used it is a cost benefit analysis. In most cases system effectiveness (or IT) evaluations are performed in qualitative terms, not in quantitative. It is interesting to remark that management decisions with regard to the implementation of MRSs are as a result based on qualitative observations rather than quantitative measurements.

The business objectives and perceived success/risk factors are summarised per system in table 5.
 

Table 5. Summary table business objectives, and success and risk factors.

© 1995-2002 Martijn Hoogeveen