Agent-mediated Electronic Commerce

1
Agent-mediatedElectronic Commerce

• introduction

Luk Stoops programming laboratory VUB
2
Consumer Buying Behavior Model

• Need Identification
• Product Brokering
• Merchant brokering
• Negotiation
• Purchase and Delivery
• Product Service and Evaluation

3
Agent Systems
4
Need Identification

• Becoming aware of unmet need
• Stimulating trough product information
• Problem Recognition
• (Engel-Blackwell model)
• Agents
• alternate publicity
• personalized publicity
• ad busters

5
(No Transcript)
6
(No Transcript)
7
(No Transcript)
8
Product Brokering

• What to buy
• Critical evaluation of retrieved product
  information
• Agents
• allow shoppers to specify constraints on a
  products features
• feature filtering
• recommend products via word of mouth
• collaborative filtering

9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
Merchant brokering

• Where to buy
• BargainFinder
• request price from 9 merchant Web sites
• 1/3 blocked all of his requests
• Jango
• request originated from consumers browser
• Kasbah
• distributed trust and reputation mechanism

15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
Negotiation

• Auctions on the web
• eBay
• On Sale
• Yahoo
• gt90 active online auctions
• Business-to-business transactions
• Fastparts (semiconductor)
• FairMarket (computer)

19
Auctions

• Hostile characteristics
• first-price open-cry
• winning bid gt market valuation

• Short term benefit
• long-term detriment

20
Auctions Disadvantages

• Bids are non-retractable
• Products are non-returnable
• Long delay between
• negotiation
• Purchase and delivery
• Only the highest bidder(s) can purchase
• Shills !
• Buyer coalitions !

21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
AuctionBot

• General purpose Internet Auction server
• University of Michigan
• Start a new auction
• Bid in an existing auction.
• Facilities for
• examining ongoing auctions
• inspecting your own account activity
• Free of charge.

30
(No Transcript)
31
Kasbah Buying Agents

• Product description
• Minimum price
• Maximum price
• Best price so far
• Time constraints
• Report activities
• Product condition
• Locality

• Minimum reputation
• Horrible
• Difficult
• Average
• Good
• Great
• Strategy

32
Buying Agents Strategies
33
Kasbah Selling Agents

• Product description
• Initial price
• Lowest price
• Time constraints
• Report activities
• Product condition
• Locality

• Minimum reputation
• Horrible
• Difficult
• Average
• Good
• Great
• Strategy

34
Kasbah Find Agents

• Monitor market for specific products
• timespan
• price domain
• Buying agents monitor
• Selling agents monitor

35
Generic - Comparative
36
(No Transcript)
37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40
(No Transcript)
41
(No Transcript)
42
(No Transcript)
43
(No Transcript)
44
Purchase and Delivery

• Security agents
• Agents monitoring
• Production
• Delivery

45
Tete-_at_-Tete (MIT Media Lab)

• Negotiates across multiple terms
• warranty length and options
• shipping time and cost
• service contract
• return policy
• quantity
• accessories
• payment options
• loan options

46
(No Transcript)
47
(No Transcript)
48
Reputation systems
49
Beginners Reputation

• It is relatively easy to adopt a new
• or change one's identity.
• If a user ends up having a reputation value lower
  than the reputation of a beginner, he would have
  an incentive to discard his initial identity and
  start from the beginning.
• Desirable that while a user's reputation value
  may decrease after a transaction, it will never
  fall below a beginner's value.

50
Reputation Improving Rate

• Even if a user starts receiving very low
  reputation ratings, he can improve his status
  later at almost the same rate as a beginner.
• If reputation the arithmetic average of the
  ratings received since the user joined the
  system users who perform relatively poorly in
  the beginning adopt a new identity to get rid of
  their bad reputation history.

51
Fake Transactions

• Two friends might decide to perform some dozens
  of fake transactions, rating each other with
  perfect scores so as to both increase their
  reputation value.
• Even if we allow each user to rate another only
  once, another way to falsely increase one's
  reputation would be to create fake identities and
  have each one of those rate the user's real
  identity with perfect scores.

52
Desiderata Reputation Systems

• Ratings given by users with an established high
  reputation in the system should be weighted more
  than the ratings given by beginners or users with
  low reputations.
• Reputation values of the users should not be
  allowed to increase at infinitum
• eBay a seller may cheat 20 of the time but he
  can still maintain a monotonically increasing
  reputation value.

53
System Memory

• The larger the number of ratings used in the
  evaluation of reputation values the highest the
  predictability of the mechanism it gets.
• However, since the reputation values are
  associated with human individuals and humans
  change their behavior over time it is desirable
  to disregard very old ratings.

54
Sporas

• New user minimum reputation
• Reputation never under that minimum
• Ratings after each transaction
• Two users may rate each other only once
• Users with high reputation experience much
  smaller rating changes

55
SporasReputation Evolution
56
Trusting Friends of Friends(Histos)
57
- Value of other user- Weight received (older
version)
58
- Value of other user- Weight received
59
- Value of the two users receiving an average
rating
60
- Value of the user rated- Weight received
(Rater has 1500)
61
Product Service and Evaluation

• Agent based
• Distributed Reputation mechanism
• Distributed trust mechanism
• Collaborative rating among the consumers
• Personalized evaluation of the various ratings
  assigned to each consumer or merchant

62
(No Transcript)
63
Recommender Systems

• Content-based filtering
• keyword-based
• extracting semantic information
• Collaborative-based filtering
• consumers ranking
• Constraint-based filtering
• constraint satisfaction problem (CSP)
• scheduling - planning - configuration

64
Me
65
Conclusions a New Game

• More fair prices (in 87 lower, EY study)
• Increased efficiency
• First movers are long-term winners
• Not playing losing
• Brands less important
• Knowing the customer owning him

66
Literature

• Agent-mediated Electronic Commerce A Survey
  Robert H. Guttmann, Alexandros G. Moukas, Pattie
  Maes
• Collaborative Reputation Mechanisms in Electronic
  Marketplaces
• Giorgos Zacharia, Alexandros Moukas, Pattie Maes
• http//ecommerce.media.mit.edu
• http//www.personalogic.com
• http//www.firefly.com
• http//www.jango.com
• http//kasbah.media.mit.edu
• http//www.ebay.com/aw
• http//auction.eecs.umich.edu
• http//ecommerce.media.mit.edu/tete-a-tete