A Quantitative Analysis of the User Behavior of a Large E-Broker Virg´ılio Almeida Wagner Meira Jr. Victor Ribeiro Nivio Ziviani Dept. of Computer Science Miner Technology Group Univ. Federal de Minas Gerais, Brazil Belo Horizonte, Brazil virgilio, meira, nivio @dcc.ufmg.br victor @miner.com.br Abstract The Internet and the World Wide Web provide a global virtual marketplace. However, there is little information about the behavior of e-commerce users worldwide. The goal of the paper is twofold. First, we give an overview of the architecture and implementation of the Miner Family of Web Agents for e-commerce. Then, we present a quan- titative study of the user behavior of a large e-broker (i.e., the BookMiner). Considering that the e-broker is used by a large number of users that only speak Portuguese and live in Brazil, we discuss the influence of regional and cul- tural issues on the e-commerce activities. Although the Web opens a company to a global market, our findings clearly indicate that e-commerce is strongly tied to regional issues, such as language, national customs and regulations, cur- rency conversion and logistics. Also, the Internet infras- tructure, mainly the intercontinental links, affects the user behavior. 1 Introduction Web-based electronic markets are adequate for information-based products (e.g., news, software, fi- nancial services, ticketing services) and also for order retailing of some non-digital products such as books, CDs, flowers, cars, among others. More and more companies around the world are creating e-commerce sites that support lists of products and/or services, price information, and commercial transactions. As a consequence, the amount of available information and the number of potential customers in the Web is growing very rapidly [13]. This work has been partially supported by Project SIAM/DCC/UFMG, grant MCT/FINEP/PRONEX number 76.97.1016.00, CNPq grant 520916/94-8 (Nivio Ziviani), CNPq grant 300437/87-0 (Virg´ılio A.F. Almeida) and CNPq grant 380134/97-7 (Wagner Meira Jr.) Though useful information may exist somewhere, it is not always easy to find what a user is looking for on the Web. Since the Web is large and growing exponentially, it is impractical to exhaustively browse the Web looking for products and services. Therefore, one of the biggest challenges faced by electronic customers is the information overload, that hampers the growth of the online buying pro- cess. Although there are several different models for rep- resenting e-customer behavior, there exist some basic steps that are shared by most models [9], such as: need identi- fication, product search, merchant search, negotiation, pur- chase and delivery, and product service and evaluation. In order to boost e-commerce activities, tools and services are needed to help customers in each of these basic steps. Two classes of services have been particularly useful for Web customers: (i) search engines and directories, and (ii) elec- tronic brokers (e-brokers). Large-scale search engines try to be comprehensive and any search usually returns many of related and unrelated information as a result of a user query. Most search en- gines are robot-based and index the whole Web as a full- text database. Robots [10, 12, 17] are software programs that traverse the Web collecting new or updated pages and sending them to a server where they are indexed. The in- dex is used to answer queries submitted from anywhere in the Internet. According to a recent study [13], the number of Web pages is estimated from 200 to 320 million. The same study says that the largest search engines in Web cov- erage are Hotbot, AltaVista, Northern Light, Excite, Infos- eek, and Lycos, and the percentage of the Web indexed by those search services varies from 3% (Lycos) to 34% (Hot- bot). A different study [6] says that the coverage of the Web by the larger search engines ranges from 28% to 55%. The main problems of search engines are high volume of data, recollection of data because of the dynamic nature of the Web, saturated communication links, and overloaded Web servers. On the other hand, high quality human maintained di-