Document Sample
23 Powered By Docstoc
					   Authoring and Publishing of Online Business-to-
    Business Electronic Catalogs With Parametric
                 Search Capabilities
                             Danish, Adel;
                         STANDARDATA Egypt
    13 Ahmed Oraby St. Mohandessin Cairo Egypt; adanish@sdata.egnet.net

     Abstract: Business electronic catalogs are used by professionals who need accurate and
     detailed information about the products they are looking for. This paper offers a
     comparison between different online catalog search methods and an assessment of the
     results obtained by the parametric search method in some commercial sites in the USA.

1. Objectives

Most of the Electronic Catalogs today on the Internet are static HTML pages with elementary
search facilities (index, keyword, navigation through a maze of menu trees) that do not offer
the user neither the required detailed information nor the appropriate search tools.
Furthermore, because the information is not organized in databases it is extremely costly to
update and to translate in other languages.
Business buyers do not “surf” the Internet: they are on a paid mission, under time pressure,
looking for a specific product. They differ from the consumer shopper in that they need
detailed information about that product (data sheets, detailed specifications, CAD drawings).
They also need a fast, accurate and user friendly search tool.
All catalog searches are driven by a search tool, a mechanism that enables a user to find what
he or she is looking for. In most business-to-business commerce, the objective of the search is
to find the model number or part number that satisfies a set of selection criteria.
When designing an online catalog, a clear positioning of the potential users of that catalog is
necessary. The design should take into consideration the most suitable search tools. As an
example in Figure 1 the consumer looking for a flower arrangement to send to his wife for
their 10th wedding anniversary does not need a parametric search to select the part number of
the flower! However, this same consumer, when buying a battery for his cellular phone over
the Internet from the JC-Penny Catalog, needs a search engine that would help him to identify
rapidly, accurately and easily the part number of the “thin battery pack and charger” he needs
to order for his “Alcatel model HC-800”.
When publishing a catalog like the JC Penny catalog on the Internet, the paper catalog has to
be completely redesigned and rethought. As an example, the current paper catalog includes a
very limited number of VCR models. This is not only due to the high cost of publishing a 4-
color paper catalog, but also to avoid confusing the buyer with too many models. On the
Internet things are different; users could have a wide selection of VCR’s, provided that the
catalog offers an easy to use search tool that could help them select a VCR according to their
selection criteria and to compare features between the different models.

                 JC Penney                                            AMP
                 LL Bean                                               HP

 Size                                     NECX

                  Flower                                         ACME
                  Shop                                          Drill Bit
             Consumer                                         Business
         Figure 1 Business-to-Business vs. Business-to-Consumer Electronic Commerce

2. Methods :
There are four electronic search methods currently in use for Electronic Catalogs:

2.1 List Method
In the List Method, the user is presented with a list of model numbers or part numbers; for
example, the list of notebook model numbers. The difference between the different model
numbers is not shown, but each model number is hyperlinked to a datasheet. The user clicks
on a product number to view its data sheet. If the data sheet does not meet the user’s
requirements, then he or she must view another data sheet, and so on.

The drawback of this method is that the user has to view each datasheet separately and either
print each one or memorize its content in order to compare between products. The higher the
number of models in the list, the more difficult it is for the user to make comparisons. This
method does not necessarily require the use of a database.

2.2 Menu Tree
In this search method, the user is presented with a succession of menus. For example, if the
user is looking for a notebook computer, the first menu might offer the choice of active matrix
versus passive matrix. After the user makes a selection, a second menu is displayed. For
example, the choice might be 800 Megabyte disk or 1.3 Gigabyte disk. This process
continues until the user reaches a single model number or part number. Then the user can
view the datasheet of this particular model number or part number.
The drawback of this method is that the sequence of the menus is predetermined. So, if the
user is looking for a notebook with a 1.3 Gigabyte drive and does not care whether it is active
matrix or passive matrix, he or she will still have to decide between the active vs. passive
option first. If the user picks, say, passive matrix just to get to the next menu, and there are no
passive matrix notebooks with 1.3 Gigabyte drive, the next menu will not show 1.3 Gigabyte
as an option. So, the user will have to go back to the first menu and choose another path until
he/she either finds what he/she wants, or finds out that it doesn't exist.
The menu tree can be created using static HTML pages without the need for a database.

2.3 Database Query Form
In this method, the user is presented with a form showing the list of available features. He or
she then selects the desired features and submits the form. The product database is queried,
and the system returns with either "no products match the selected features" or a list of
product numbers which match the selected features. The user then has to view the different
datasheets of the list in order to know the differences, just like in the List Method.
The drawback of this method is that if the user selects many features, there is a high chance of
getting a "no match", and if he or she selects few features, he or she will get a long list of
matching product numbers. This leads to the problem of distinguishing between many
different datasheets.

2.4 Parametric Search:
It is important to distinguish here between searching for a document on a particular subject
and searching for a part number in a catalog. When searching for a document, a keyword
search is appropriate, although it often results in a large number of matches that the user must
narrow down by adding additional keywords. When searching for a part number, a parametric
search is more appropriate. In a parametric search, each product feature is stored in an
independent element in the database. The user selects only one feature at a time before
submitting the form, beginning with the most desired option. The system queries the product
database and responds by showing the number of products which match the selection criteria.
The screen, meanwhile, is updated to show only the features which can be selected next. It
eliminates options that can't be found, based on the search just conducted.
The key issues involved in selecting a search tool are similar to the fundamental issues that
pertain to electronic catalog performance. These issues include the following:
   Search time - how long does it take for the system to locate a product or part that matches
    user criteria?
   Accuracy - how accurate is the tool in performing the required search? Does it display
    “false positives” and other erroneous or irrelevant information?
   Ease of use - how easy to use is the tool, measured in terms of time and degree of
    difficulty required to sort through the displayed information in order to find a satisfactory
    answer? How customizable is the tool in meeting user priorities?

As shown in Table 1, search methods differ significantly as to their effectiveness in helping
the user locate desired information quickly and with a minimum of frustration.

                          Keyword       List        Menu      Database    Parametric
                           Search      Method       Tree       Query       Search

       Search Time          Poor        Poor         Poor        Poor       Excellent

       Accuracy             Poor        Good        Good        Good        Excellent

       Ease of Use          Good        Good        Good        Good        Excellent

                            Table 2. Evaluation of Search Methods
Parametric search catalogs require a detailed analysis of a product family in order to identify
the search parameters (features) and the values of these features for each product in the
catalog. The features and their values for each item are then entered in a relational database
specially structured for a parametric search.
A good example of Parametric Search implementation is “Step Search” from Saqqara Systems
Inc.[Danish96]. A key advantage of Step Search is that the user never gets a "no match," since
the available features are updated after each selection. Also, the user can select the features in
the most important order, so he or she can establish a priority starting with the most important
feature. Each search is essentially customized to the user. Another advantage of Step Search is
that it provides help information on the different features in order to guide the user in the
selection process. Finally, when the user reaches a reasonable number of matches (fewer than
10), he or she can view a comparison table which shows the differences among the remaining

3. Results
Parametric Search is currently used in some of the largest catalogs on the Internet. One of the
most successful sites using this technology is the AMP Connect site[AMP]. AMP is a
worldwide leader manufacturer of Electronic Connectors with a 5.2 Billion dollars annual
revenue. AMP launched it’s parametric search catalog in January 1996 with more than 65,000
parts fully described online. The catalog receives around 35,000 hits per day. 200 Repeat users
visit the catalog daily with a monthly increase of 15%. There are 32,000 registered users from
100 countries and 200 new users register every day.
Other Fortune 500 companies like Hewlett Packard are also using the same technology. After
a pilot project for HP power supplies[HPb], HP has decided to put more product catalogs
online using the parametric search technology[HPa].
Parametric search is not limited to use of large companies. A company by the name of 800-
Batteries carries more than 7,800 replacement batteries for laptops, cellular phones and
camcorder products is using parametric search to allow its potential customers to find and
order the right battery part number in a couple of keystrokes.
One of the major benefits of parametric Search tools is the collection of marketing
information. As it was mentioned before, when using a parametric search the user selects one
feature at a time beginning with the most desired feature. By logging these transactions, the
catalog owner can sense online what are the most desired features for a certain product and
emphasize them in his marketing campaign. Furthermore, the transaction analysis tool could
also detect what are the feature combinations that are most selected. This analysis could allow
the product manufacturer to have an unprecedented understanding of what the market wants.
Needless to say that such analysis could be sorted and classified by geographic region (using
the caller IP address) to achieve even more accurate research analysis.
That sort of detailed and accurate market analysis was until now limited to the large
organizations that could afford paying for such an expensive market research . Today, a
company like 1-800-Batteries, which is more of a small to medium size enterprise can have
access to that sort of information instantaneously at almost no cost.
To conclude, parametric search engines offer manufacturers, vendors and users new tools for
electronic commerce. Catalogs based on this tool are well suited for both business-to-business
and business-to-consumer electronic commerce. Parametric search engines are best suited for
online catalogs with large number of products to select from. Byproducts of parametric search
like market needs analysis (through transaction logs monitoring) makes it even more attractive
to use.

[AMPa]              AMP Catalog homepage (connect.amp.com)
[Andrews96a]        Whit Andrews: Better Searching, In (Not) So Many Words, Web Week,
                    June 3, 1996
[Batteriesa]        800-Batteries Catalog (sphinx.saqqara.com/cgi-
[Bock95a]           Geoffry Bock: Step Search from Saqqara Systems: Interactive browsing
                    for Electronic Commerce, Patricia Seybold Group, Workgroup
                    Computing Report, Vol. 18, No. 12, December 1995
[CommerceNeta]      CommerceNet homepage (www.commerce.net)
[Danish96a]       S. Danish: How to Build a Successful Business-to-Business Online
                  Catalog, Capitalizing on the Promise of the Internet. SAQQARA
                  Systems, Inc., 1996
[DM96a]           Esther Dyson and Jerry Michalski: Saqqara Systems: Take Your Pick,
                  Release 1.0, March 17, 1996
[HPa]             Hewlett Packard, Electronic Components Catalog
[HPb]             Hewlett Packard ,Test & Measurement Catalog
[Fischer96a]      Layna Fischer: Electronic Commerce: Profiting From Business Online,
                  Future Strategies, Inc. Book Division, 1996
[Loshin96a]       Pete Loshin: The Electronic Marketplace: Transforming the Way We Do
                  Business, PC Today, July 1996
[McNaughton96a] Kora McNaughton: Fifty Hot Internet Companies, Upside, April, 1996
[SWB95a]          A. Segev, D. Wan, and C. Beam: Designing Electronic Commerce For
                  Business Value - A CommerceNet Pilot Project Report, The Fisher
                  Center for Information Technology and Management, Haas School of
                  Business, University of California, Berkeley, 1995

Shared By: