Software agents: are programs to which one can delegate (aspects of) a task they differ from 'traditional"

software in that they are personalized, continuously running and semi-autonomous. These qualities make agent useful for a wide variety of information and process management tasks. E-Commerce encompasses a broad range of issues including security, trust, reputation, law, payment mechanisms, electronic product catalogs (EPC), intermediaries, multimedia, shopping experiences, and back-office management. Agent technologies are desirable in E-Commerce areas. Agent technologies employ in E-Commerce: The more time and money that can be saved through automation. The easier it is to express preferences. The lesser the risks of making sub-optimal transaction decisions.

Explore the roles of agents in E-commerce in the context of a common model (CBB). CBB (Consumer Buying Behavior) model research and comprises the actions and decisions involved in buying and using goods and services. CBB model is a powerful tool to help us understand the roles of agents as mediators in E-commerce. CBB research focuses on the retail market. Its concepts pertain to business-to-business and consumer-to-consumer markets as well.

There are several models and theories that attempt to capture consumer buying behavior (CBB). All these models share list of six fundamental stages guiding CBB model. This six stages show where agent technologies apply to the consumer shopping experience and allows us to categorize the existing agent-mediated E-commerce system. CBB (Consumer Buying Behavior) six stages: 1. 2. 3. 4. 5. 6. Need Identification. Product Brokering. Merchant Brokering. Negotiation. Purchase and Delivery. Product Service and Evaluation.

1-Need Identification:
This stage characterizes the consumer becoming aware of some unmet need. Within this stage, the consumer can be stimulated through product information.

2- Product Brokering:
This stage includes the retrieval of information to determine what to buy. This includes the evaluation of product alternatives based on consumer criteria.

3-Merchant Brokering:
This stage combines the consideration set from the previous stage with merchant specific information to help determine who to buy from. This includes the evaluation of merchant alternatives based on consumer criteria.

4-Negotiation:
This stage is about how to determine the terms of the transaction. Negotiation varies in duration and complexity depending on the market. In traditional retail market, prices and other aspects of transactions are fixed no room for negotiation. In other markets, the negotiation of price or other aspects of the deal are needed to product and merchant brokering.

5- Purchase and Delivery:

The purchase and delivery of a product can be signal the termination of the negotiation stage or occur sometimes afterwards.

6-Product Service and Evaluation

This stage involves product service, customer service, and an evaluation of the satisfaction of the overall buying experience and decision.

These stages represent an approximation and simplification of complex behavior. CBB stages often overlap and migration from one to another can be non-linear. From those, we can identify the roles of agents in E-commerce. The nature of agents make them well suited for consumer behaviors involving information filtering and retrieval, personalized evaluations, complex coordination and time based transactions. Specifically these roles most notably to the Product Brokering, Merchant Brokering, and Negotiation stages of the Consumer Buying Behavior model.

The Product Brokering stage of the CBB model is where consumers determine what to buy. This occurs after a need has been identified (i.e., in the Need Identification stage) and is achieved through a critical evaluation of retrieved product information. Table l shows several agent systems that lower consumers' search costs when deciding which products best meet their needs: personaLogic, Firefly and Tete-aTete. PenonaLogic is a tool that enables consumer to narrow down the products that best meet their needs by guiding them through a large product feature space. The system filter out unwanted products within a given domain by allowing shopper to specify constraints on a product's features. A constraint satisfaction problem (CSP) engine

then returns an ordered list of only those products that satisfy all of the hard constraints.

Firefly services help consumer find products. However, instead of filtering products based on features, Firefly recommends products via a "word of mouth" recommendation mechanism called automated collaborative filtering (ACF). ACF first compares a shoppers product ratings with those of other shoppers. After identifying the shopper's "nearest neighbors" (i.e., user with similar tastes), ACF recommends products that they rated highly but which the shopper may not yet have rated.

This stage combines the consideration set from the previous stage with merchant specific information to help determine who to buy from. Whereas the Product Brokering stage compares product alternatives, the Merchant Brokering stage compares merchant alternatives. Three representative agent systems that perform merchant brokering are BargainFinder, Kasbah and Jango BargainFinder was the first shopping agent for on-line price comparisons. Given a specific product, BargainFinder requests its price from each of nine different merchant web sites using the same requests as a form of web browser. It displays the result in a list ready to be compared from the consumer. Merchants dont want to compete on price only so they blocked all price requests from BargainFinder. Value added services that merchants offered in their web site were being bypassed by BargainFinder and therefore not considered in the consumers buying decisions. In short, companies competing on price only wanted to be included in the BargainFinder, but the other didnt want.

Jango can be viewed as an advanced BargainFinder, The original Jango version solved the merchant blocking issue by having the product requests from each consumers web browser instead of from a central site as in BargainFinder. This way, requests to merchants from a Jango-augmented web browser appeared as requests from real customers. Once a shopper has identified a specific product, Jango can simultaneously query merchant sites (from a list now maintained by Excite, Inc.) for its price. These results allow a consumer to compare merchant offerings on price.

Kasbah is an on-line, multi-agent system. A user wanting to buy or sell a good creates an agent, gives it some strategic direction, and sends it off into a centralized agent marketplace. Kasbah agents pro-actively seek out potential buyers or sellers and negotiate with them. Each agent's goal is to complete a deal subject to a set of user-specified constraints, such as a desired price, a highest (or lowest) acceptable price, and a date by which to complete the transaction.

This stage is about how to determine the terms of the transaction. The Negotiation stage is where the price or other terms of the transaction are settled. Three representative agent systems that employ differing negotiation protocols are Kasbah, Tete-a-Tete and AuctionBot.

Negotiation in Kasbah is straightforward. After buying agents and selling agents are matched, the only valid action in the negotiation protocol is for buying agents to offer a bid to selling agents with no restrictions on time or price. Selling agents respond with either a binding yes or no. AuctionBot is a general purpose Internet auction server. AuctionBot users create new auctions to sell products by choosing from a selection of auction types and then specifying its parameters, such as auction time, reserved price, etc. Once the auction is created the sellers negotiation is completely automated by the system as defined by the auctioneer protocols and parameters of the selected auction.

Tete-a-Tete provides a unique negotiation approach to retail sales. Unlike most other on-line negotiation systems which competitively negotiate over price, Tete-a-Tete agents cooperatively negotiate across multiple terms of a transaction e.g.. Warranties, delivery times, service contracts, return policies, loan options, gift services, and other merchant value-added services.

This negotiation takes the form of multi-agent. Tete-a-Tete's shopping agents use the evaluation constraints captured during the Product Brokering and Merchant Brokering stages as dimensions of a multi-attribute utility. Tete-a-Tete integrates all three of the Product Brokering, Merchant Brokering, and Negotiation CBB stages.

Most of the technologies that support Agent-Mediated either directly or indirectly from Artificial Intelligence. Such as: Cognitive Models To extract meaning of ambiguous web pages Planning trips Learning users Music preferences Deciding which car to buy? The majority of product recommender systems are developed using: Content based Collaborative based Constraint based

1-Content based filtering:

The system processes information from various sources and tries to extract useful elements about its content. Techniques Used: can Vary Greatly in complexity Such as: Keyboard-based search simplest techniques involves matching different combinations of keywords. Extracting semantic of the content by using techniques like associative networks of keywords in a sentence, or a directed graph of words that form the sentence.

Examples of systems that use Content

based filtering:

Systems like BargainFinder, Jango try to collect information from many different web information sources .and different sources have different input method, presentation methods. So the recommender systems must adjust their interaction methods depending on the web site. They used wrappers to transform the information into a common format. BargainFinder Hand-coded by Andersen Consulting programmers. was good at the beginning then it became very hard to scale and it require maintenance of the wrapper for each site if the access method changed or its catalog changed Jango method is not perfect but boasts a nearly 50% success. Jango Automatically creates wrappers by generalizing from example query response to online merchant DB.

2-Collaborative Filtering:
Firefly uses Collaborative Filtering technology to recommend product to customer. System using collaborative technique use feedback and ratings from different consumers to filter information. It doesnt attempt to analyze or understand the feature or the description of the product, all its concerns about its rank.

3-Constraint Based:
Uses the feature of items to determine the relevance. It requires problem and solution formulated in terms of variables domains and constraints.

Once formulated in this way a number of general and powerful constraint satisfaction problem techniques can find a solution.

Has 3 types Finite Domain: Finite set of variables Finite set of constraints : define the relationship among variables In personaLogic CSP techniques are used in the Product Brokering CBB stage to evaluate product alternatives CSP techniques are core to Tete-a-Tete and are used to assist shoppers in the Product Brokering, and Negotiation CBB stage. This achieved by providing product constraints such as: Price Delivery Time Warranty

Traditional shopping experiences can be quite diverse depending on the needs of the customer for instances: Consumer may just browsing without real intention to buy any thing Know what to buy but not familiar with its specs Missing positive feedback They care about trust especially when money involved Customers preferred simple predictable agent with pre-determined negotiation strategies

Its a form of decision making where two or more parties search for possible solutions with the goal to reach a consensus. Has 2 general approaches: Distributive negotiation focus on how each self-interested party can take the largest piece of the available pie. Example->AuctionBot and Kasbah. Integrative negotiation focus on all of the terms of a transaction and the parties involved try to enlarge the available Pie. The benefit of this approach over the Distributive is that its possible to find new points in the solution space that benefit the both buyer and seller. Ex Carnegie Mellon University which finds its roots in Distributed artificial intelligence Another approach is Contract Net.

Since there are a lot of agents and they use different protocols there are Knowledge Interchange Format, Knowledge Query Manipulation Language, Ontolingua. These were designed for different agents so they could describe knowledge and communicate it meaningfully in order to interoperate. In Business-to-Business world there is EDI ( Electronic Data Interchange) its a set of ANSI Protocols Other E-Commerce protocol my need to include Internet-based EDI (EDINT) , XML/EDI , Open Buying on the Internet (OBI) for Highvolume , Low-dollar business-to-business purchase .

 Also another protocols such as Open Financial Exchange(OFX) for financial transactions ,Secure Electronic Transaction (SET) for credit card transactions, and Open Profiling Standard (OPS) and Personal Privacy Preferences Project (P3P) for defining privacy for consumer profile data . EDI (Electronic Data Interchange) facilitate large-scale, repetitive, pre-arranged transactions between businesses in specific industries with each industry adapting the EDI protocol to its specific needs. Standard EDI protocol are performed through expensive proprietary Value Added Network (VAN)

EDI Disadvantages:
Ambiguous Expensive to implement and maintain used Focused on large scale business-to-business transaction