An Intelligent Tool for All-to-All Sales on the Internet
Platform for Compilation of Commercial Offers and Requests for Companies and
Customers
Radoslav Fasuga, Pavel Stoklasa and Martin Němec
Department of Computer Science, VŠB Technical University of Ostrava, 17.listpadu 15, Ostrava-Poruba, Czech Republic
Keywords: Offer Portal, Intelligent Search Engine, Parametric Search, Electronic Commerce, E-Shop, Second Hand,
Offers, Requests, Exchange, Multilingual Implementation, Multimedia System, Product Catalogue.
Abstract: This paper deals with the issues of building an intelligent sales portal. The portal features include a price
comparator, entering of individual offerings and requests for concrete and specific products and services.
Product offers can be entered by choosing from predefined product options and specifications, or as new
specifications. The paper describes matching of sales for new products, services and used goods. The design
philosophy combines the features of product aggregators, advertising and second-hand sale systems and
auctions with so-called fair direct offers in both public and anonymous offer and request proceedings. The
implementation is designed for both companies presenting their products, services and requests, and ordinary
customers and their individual offers and requests. The core system component is the intelligent search and
offer/request matching process. The identical product-to-product matching model and searching based on
product similarity is described. A model ideal product and most similar offer can be defined. The paper
includes a discussion of the issue of working with a large quantity of highly structured data, their pre-
processing and effective visualisation. The conclusion outlines the project ambitions in the area of building a
European offer and request platform among business and customer clients.
1 INTRODUCTION
Monitoring of offers and requests on the Internet
plays an increasingly significant role today.
Customers including individuals, households, small,
medium, large and multinational companies require a
high degree of customisation of supply and demand,
seeking new business opportunities and consolidation
of existing sales and business relations.
The objective is to minimise the time required for
compiling a sales offer or request and effective
acquisition of a relevant response leading to an actual
sale. The paper discusses existing solutions and
compares their advantages and drawbacks. It reflects
user knowledge of products or services offered and
requested. It outlines customer decision-making
models, their preferences, functional and general
requirements, the importance of pricing, (regional or
global) availability, delivery method, warranty
conditions, servicing and extra services (Wang and
Zhang, 2012).
This paper deals with the issue of effective
presentation of products and services, definition of
sales offers and requests, collection of the relevant
response, evaluation and support of the actual sales
process. In addition, it describes methods of
evaluation of relevance and credibility of trade
partners for further decision-making (Fasuga et al.,
2014).
It discusses both matching algorithms and
technical resources necessary for implementation of
such an extensive portal with finely structured data
and a high degree of diversity of product offers. It
describes methods of minimal, partial and complete
pre-processing of sales offers and requests with
reference to additional advanced searching and
matching methods.
Moreover, the article describes concrete
implementation of a real-world portal dealing with
the issues at hand. It discusses opportunities for
further expansion and establishment of a European
all-to-all (A2A) sales and offering platform (All =
businesses, customers, governments) (Figure 1).
263
Fasuga R., Stoklasa P. and N
ˇ
emec M..
An Intelligent Tool for All-to-All Sales on the Internet - Platform for Compilation of Commercial Offers and Requests for Companies and Customers.
DOI: 10.5220/0005536202630270
In Proceedings of the 12th International Conference on e-Business (ICE-B-2015), pages 263-270
ISBN: 978-989-758-113-7
Copyright
c
2015 SCITEPRESS (Science and Technology Publications, Lda.)
Figure 1: All-to-All diagram for subject interaction.
2 INSPIRATION DRAWN FROM
EXISTING SOLUTIONS
Existing sales methods on the Internet are based on
established models of supply and responding
demand. The aim is to elicit the buyer's interest in the
product, or to boost their decision in our favour.
Companies or individuals offer their products and
services on their own websites, product aggregators,
paid advertising campaigns, social networks, auction
portals, and one-to-one communication with
registered users and previous users. Customers
usually wait for dealers' offerings, based on which
they then make decisions.
Figure 2: Sales concepts integrated into single system.
However, there are also different sales models,
which are equally effective, reduce costs of
promotion and only collect relevant customer (buyer)
requirements. These are advertisement and second-
hand portals, goods or service exchange solutions,
and requesting systems for (typically large) contracts.
In addition to conventional offering, these also
feature an equal counterpart in the form of more or
less specific requests. The advantage is the cost
minimisation on the part of the entrant (buyer or
seller), who only specifies their requirements and
waits for relevant replies and offers from trade
partners (Figure 2) (Cai et al., 2013).
2.1 Product Price Comparators
Active (knowledgeable = decided) users search for
specific products and services in price comparators
increasingly often (example: Google Merchant
Products). Companies place their offers in these
systems and update them on a regular basis. They
offer concrete (matched) products at a declared price
and availability. Full-text search methods are then
used for other products.
The disadvantage of this design is that the
offering company does not have direct information
about specific requirements and price expectations of
specific customers. It can therefore only respond to
competing offers, not to customer requirements.
Customers can comment on products, their own
experience and satisfaction with them, as well as with
companies offering the products. There may be
official reviews, detailed descriptions and classified
properties of the products. Classified properties are
used for detailed searches. The user may specify
technical parameters of a product and the most similar
products are found.
2.2 Auction Systems
Auction portals (eBay, Aukro, etc.) represent a large
portion of the direct online sales market. These
portals were originally focused on selling used goods,
worse quality goods, clearance sales and closing
sales. Today, they are used as a secondary market for
conventional e-shops and sales portals, or as a
primary market for small sellers. Again, the sales are
only oriented at offering specific products or services.
Potential buyers bid up on each product and it is then
sold to the highest bidder (Fasuga et al., 2010).
The auction environment is a fast alternative to
conventional sales. Its chief benefit is the seeking of
potential customers and interest groups without any
additional costs. The drawbacks for the seller are the
limitation to selling to the highest bidder only, long
periods in between sales, and the impossibility to
respond to other, lower bids.
For the buyer, there is a single winner in a bidding
auction (unless it concerns a sale of multiple items).
The drawback is the impossibility to respond to
competing offers and the sale end time being set by
the seller, not the buyer. It is therefore more like a
game with a random winner than a real-world auction
process respecting both sellers and buyers.
Customers
Govern‐
ments
Businesses
Globaltradesystem
Product
price
compar
ator
Auctionsystem
Debet Credit
Discoun
tportal
Avertisingsystem
Second‐
hand
Adverti
sement
ICE-B2015-InternationalConferenceone-Business
264
2.3 Discount Portals
Discount portals are another segment of the market
supporting sales of individual products. Here,
companies offer selected bargains in limited
quantities under advantageous conditions. There are
many discount portals and even discount offer
aggregators exist today. Companies can thus make
offers of only limited quantities of products on offer
or generic discount vouchers for their entire product
range, but then they lose control over which products
will be marked down in the end.
Customers are then limited by specific offers or
purchases of inexpedient (unneeded) product and
service combinations (Fasuga et al., 2011).
2.4 Advertising and Second-Hand
Portals
Advertising newspapers, magazines and second-hand
offers are still popular chiefly for private offers and
requests. Most companies have shifted from print
periodicals to the Internet and they now offer portal
solutions, which collect general offers and requests.
Users select their area of interest, then specify
their requirements in a textual form, which they can
complement with multimedia information (photos,
videos and more). The offers are so general that it is
difficult to match the ads withsearchers'
requirements. Reading and sifting through large
quantities of ads is time-consuming and, ultimately,
inefficient. There are no templates for typical
products and services in the category and no precisely
defined sets of monitored properties. The focus of the
second-hand solutions is mostly on used goods, worse
quality goods, goods in use or goods in private
ownership. There are also clearance sales, with the
attribute reducing the value and quality of the goods
in the buyer's eyes.
3 SALES OFFER AND REQUEST
PRESENTATION FORMAT
It is based on usual product categorisation and so-
called generic design for creating sets of descriptive
attributes (Nguyen et al., 2011) (Figure 3).
The system has to be divided into categories
(supertypes) and subcategories (subtypes). Subtypes
are specialisations of supertypes; they specify and
divide their properties and services. A list of general,
shared properties for each product class is assigned to
each supertype. Other specific properties that can be
filled in are then added by defining the subtype
hierarchy. Properties are divided into mandatory and
optional (additional). Mandatory properties always
have to be filled in; they are typically basic
classifications and divisions of offers. The generic
design structure (Diagram 1) is suitable for the
implementation; it allows dynamic addition of other
properties and their values as needed.
Figure 3: Define product properties and overlap.
3.1 Definition of Basic Properties
Moreover, there is a list of properties. Property
specifications include their name, data type, value,
unit and comments. The name of a property describes
the content stored; the same property can be stored for
multiple products of different types (e.g., screen
definition for: TV sets, monitors, tablets, notebooks,
mobile phones, cameras, etc.).
In this case, the data type specifies whether the
parameter value is entered manually, whether it is text
or a number, whether there is a choice from more
options of values.
Having these values categorised is an
unquestionable advantage. Therefore, you have a list
of permissible options when defining a product, and
only choose one or more of them. This eliminates
typing and other undesirable errors; there may be a
problem with new products and technologies, as
brand new products may have new properties,
capacity, output, etc., that are outside the normal
ranges (Kołaczkowski and Gawrysiak, 2011).
Another inseparable part of a property is its unit.
An explanatory comment, which helps clarify the
meaning of the property, can be attached to each
property. You can also comment on other variations
of the property, highlight its advantages and
shortcomings.
Basicproperties
Aditional
properties
Premium
properties
Properties
overlapping
Non‐standardized
products
AnIntelligentToolforAll-to-AllSalesontheInternet-PlatformforCompilationofCommercialOffersandRequestsfor
CompaniesandCustomers
265
3.2 Premium and Additional
Properties
Some products may have so-called premium
properties, which are not typical. Alternatively, they
may occur as new properties for the given product
(expansion of standard product features). It is
therefore necessary to include the property in the
system to make the products findable by it.
Additional properties are not directly related to a
product but its distribution, storage, transport, price,
availability, etc.
These properties are set case by case for each
seller. Their setting may depend on the product type,
dimension, weight, manufacturer, supplier, etc. The
properties have to be clearly separable from the
suppliers' official specifications.
3.3 Property Overlap
In private sales and sales of used goods in particular,
official properties may overlap, meaning a
modification of the catalogue specification (e.g., a
computer memory upgrade or hard drive replacement,
incomplete product accessories, transfer of a holiday
vacancy, etc.).
In such cases, the official description parameters
have to be replaced with new values. No new
parameters are added; only the existing specification
is modified (Kotha and Basu 2011).
The advantage of this approach is the unification
and comparability of products, efficient searching
based on defined parameters and their values.
3.4 Definition of Own
Non-Standardised Product
It is obvious that the system will not contain
specifications of all (older versions of) products and
their alternatives. It is therefore possible to insert in
the system a general offer or request, which will
respect the structure and parameters of a given
category and will be integrated into the detailed
search.
Alternatively, it can be described only briefly, as
an advertisement in the form of running text. This
then allows full-text searching only (Lahaie, 2006).
The goal of the project is to minimise the
proportion of non-standard offers and present
relevant results only.
4 DATABASE
IMPLEMENTATION
Such an extensive project requires efficient database
implementation and choice of technologies to use.
The implementation itself was preceded by database
server performance tests with identical hardware
configurations. The SQL (Structured Query
Language) test compared Microsoft SQL Server
2008, Oracle 11g, PostgreSQL, and MySQL.
MySQL, with adequate performance and acceptable
operating costs, was then selected for the
implementation. The Java language has been chosen
for the system module and crawler implementation;
the PHP technology with a suitable framework has
been chosen for the back office and the front-end
web. Intensive testing of available frameworks
(Laravel, Phalcon, Symfony 2, CodeIgniter, Zend 2,
Nette) is currently in progress.
4.1 Listed Product Catalogue
During the test operation and acquisition of
information from the XML (Extensible Markup
Language) feeds that e-shops publish in price
comparators (Heureka.cz, Zboží.cz, HyperInzerce.cz
are the product aggregators in the Czech Republic),
approximately 10,000,000 different active products
were identified. Each product has approximately 14
comparable properties, a long description of 2,500
characters on average, and graphic information in the
form of multimedia files with images and support
graphics with 4 images per product at 150 kBytes on
average. Approximately 6,000,000 of the products
include a link to the manufacturer's or distributor's
official website (Table 1) (Kannan et al., 2011).
Table 1: The total size of the product and property database
comprises.
Basic product information 4.50 GBytes
Detailed properties and parameters 1.40 GBytes
Indexes for search engines 2.20 GBytes
Full-text indexation (approx.) 8.00 GBytes
Images and multimedia information 6.00 TBytes
The estimated proportion of individual products
and changes of parameters in catalogue products is
approx. 15-20%.
Additional technologies to be used are 360-degree
rotation, panoramic pictures, product videos and
virtual tours at an estimated size of 10 MB per
product. The estimated number of products presented
in this way is 20-40%; the required data storage size
is then 30 TBytes.
ICE-B2015-InternationalConferenceone-Business
266
Selling prices, comments and reviews are
registered and periodically updated for each
catalogue product. Each product has 34 price offer
available on average, and the estimate is that 1-2
requests on average will be stored for each product.
Moreover, it is estimated that 5 comments at 1 kBytes
each on average will be stored for each product (Table
2).
Table 2: The expected increase in the database size.
For current product prices 5.44 GBytes
For requests 1.02 GBytes
For comments 50.00 GBytes
To reduce the requirements over the database, we
chose the method of caching (pre-generation) of final
pages with product information previews at an
average size of 200 kBytes of HTML (HyperText
Markup Language) code per product, as well as
search results for typical keywords and phrases at up
to 200 kBytes of resulting data per product.
The pre-generation of product and price
information:
If the description changes (immediately)
If a product description or comment is added
(immediately)
If the seller changes the selling price (within
approx. 2-5 hours depending on previous
frequency of changes)
If a request is added, request matching
(immediately upon entry).
The database engine used within MySQL. Three
engines were used for real operation, namely InnoDB
for structured data storage. MyISAM for storage of
the presentation layer and the full-text search, and the
Memory type for fast parametric searches.
The database storage facility was fitted with fast
SSD (Solid-state drive) discs and, for the multimedia
information, high-capacity 6 TBytes RAID
(Redundant Array of Inexpensive/Independent Disks)
Edition or Enterprise Edition discs. The RAID 6 +
Spare technology was chosen for the RAID discs,
with an EXT 4 fourth extended file system for Linux.
The system was set up with periodic weekly backups
and incremental backups of critical database
components in the meantimes.
Migration to Google Cloud is planned for the
future.
4.2 System Block Diagram
The system is divided into several basic modules. The
first is an XML Feed Eshop Crawler (FEC). This
system is implemented as a multi-thread console
application in Java with a shared memory and single
data storage in the working product database. This is
followed by an intelligent product matching module,
based on EAN (European Article Number since year
2009 International Article Number) or GTIN (Global
Trade Item Number) codes, keywords in product
names, numerical identifiers, and dynamically
changing intervals of acceptable prices. The system
also tries to identify products in clearance sales and,
increasingly often, products on offer that have been
returned to the seller, and pre-orders for products not
dispatched yet (Figure 4).
Figure 4: List of system implementation modules.
There is also a module testing the correctness of
presented data (TCPD). Based on a specific
algorithm, it crawls through the different e-shops, and
tries to identify the prices and the availability
declared in the feed in comparison with the real offer
on the seller's website. If the number of discrepancies
between prices and availability is increasing, other
products are scanned as well and when a limit
tolerance is exceeded, the system is penalised and
excluded from further searching.
Another tool is a system for converting obtained
data into a form convenient for presentation (COD-
FCP) to ordinary users. Another test of dishonest
conduct is a tool monitoring manipulation with
keywords, i.e., descriptions of offers that are related
to other products or brands and deliberately include
names of other products in their descriptions so that
they are shown when searched for. This conduct is
again regarded as unethical, therefore penalised by
blocking off the presented content.
This is followed by a module for searching in
offers (SO), both matched products and full-text
searches. Results of matched products are preferred
to full-text.
Then there is a tool for searching based on specific
parameters (SBSP), which generates detailed
Testingthe
correctnessof
presenteddata
Convertingdatainto
formfor
presentation
Searchinginoffers
Searchingbasedon
specificparameters
Administration
systemfor
corporationand
users
Managing
comments
Cooperationand
connectingoffers
Monitoringofprice
evolution
AnIntelligentToolforAll-to-AllSalesontheInternet-PlatformforCompilationofCommercialOffersandRequestsfor
CompaniesandCustomers
267
templates for querying; results are not pre-processed
and all queries are handled individually.
There is a user profile administration system for
both corporate clientele and ordinary users (AS-CO).
This involves management of identities and
registration information with identity authentication.
A system for managing comments (CM), reviews
and evaluations. It enables evaluation of companies,
sellers, buyers, as well as products or services offered.
A system for cooperation and connecting offers
(CCO) monitoring options for interlinking various
offers, such as a product seller and a servicing
organisation. The interlinking can be defined as
uthorised, unauthorised, undesirable, and banned
(blocked).
Monitoring of price evolution (MPE) for a
commodity or an entire market, with monitoring of
competing offers. Thus, the user can see not only their
products and offers but also the matched offers of
competitors. Based on that, they can respond and
adjust the price or selling terms and conditions.
4.3 Additional System Features in
Preparation
A general tool for importing complex data structures
such as specifications of cars, real estate, holidays,
etc. The tool will both respect inherent standards for
description of these objects and allow importation
from various data structures CSV (Comma-separated
values), XLS (Microsoft Office Excel), XML, JSON
(JavaScript Object Notation) and more. After that,
external data structures will be paired over internal
specifications of the sales portal.
Development of an application version for mobile
devices. The precondition is the integration of
responsive application design based on the Twitter
Bootstrap CSS (Cascading Style Sheets) Framework.
In the second phase, generation of information
sources respecting requirements of mobile devices,
particularly reduction to the amount of data
transferred, and last but not least, purely mobile
applications for the three major platforms - Android,
iOS and Windows Phone. Expansion to gaming
consoles and smart televisions is being considered.
Although the responsive design is sufficient for the
primary launch.
5 REQUEST AND OFFER
MATCHING METHODS
The primary feature is the module for matching offers
and requests (or conversely, requests and offers)
(Figure 5) (Ayanso and Karimi, 2015).
Figure 5: Offers lifecycle.
5.1 Offer Entry
We will describe the steps in entering an offer. The
entrant/company imports a list of products offered
into the system as per the XML Feed standard
(Google Merchant Feed). For fewer products or a
private offer, they enter products by selecting from a
product catalogue. Alternatively, they can modify
product parameters or create a new product, or enter
their request only as a text ad available in full-text
search results only.
Company imports are subject to authorisation, and
the form and content of input data are checked before
including the content in the system, along with a data
verity authentication.
Only the entrant's identity is verified for private
offers (Ong, 2011).
5.2 Offer Search
The system allows searching for offers by browsing
the product catalogue. Products are labelled as
company and private offers (requests). Another
search option is definition of specific parameters.
First choose a product and set the desired properties,
then suitable products are selected; parameters can be
further specified until a product matching your idea is
found (Overby and Forman, 2015).
If no identical products are found, you can choose
a so-called approximate search, where you define the
degree of tolerance and deviation direction for each
parameter. For example, you can tolerate a higher
product price, different delivery terms, or higher
capacity. Importance for parameters can also be
changed for the search. This means whether you care
more about the price or the properties of the product
or service (Zou et al., 2014).
Finally, there is the full-text search. You can run
this search if you have not found your requested
product; the search is made in indexed texts and
Offerentry
Offer
search
Requests
andcross‐
offers
Publicna
non‐public
offers
proceeding
ICE-B2015-InternationalConferenceone-Business
268
descriptions and, most importantly, in products
outside the catalogue.
If you are using the system as a registered user,
you can save all your search settings as templates for
repeated future use..
5.3 Request and Cross-Offer Entry
If you find a product you request, you can enter
negotiation with selected sellers. You can enter a
cross-offer for the product, define your requirements
for a better (lower) price, number of units required,
delivery terms, servicing terms, etc. The company can
thus respond to your specific individual request and
accept your requirements or refuse them. Requests
also specify whether the price is fixed and the selling
parameters final (fixed) or whether everything is open
to negotiation (Bauner, 2015).
If you cannot agree with any seller, you can enter
your specification as a request and wait if you are
addressed by a seller who would like to accept your
request.
Likewise, you can use this monitoring as a form
of watchdog, which can see how sellers' offers change
over time (discount, availability, etc.).
5.4 Public and Non-Public Offer
Proceedings
Mutual communication among offering and selling
entities may proceed in a public, anonymous or secret
mode. The offer (request) entrant always decides
about the degree of visibility of the communication.
The other party always has to accept the user's
settings. In the public mode, specific requests to your
offer and the whole communication (prices,
discussions, evaluations) are visible. In anonymous
communication, the dialogues are not visible; only
the price offers and numbers of units requested are
seen. Other customers can thus respond to visible
parameters and adjust their counter-offers. In secret
communication, none of the seller-buyer interaction
can be seen publicly (Pham et al., 2015).
Experience with the system operation has shown
that the anonymous communication setting is the
most effective; it makes the results of communication
visible without any comments or specific contacts.
6 PILOT SYSTEM OPERATION
The system has been integrated into the portal
kartoteka.cz. It had been developed since 2000 as a
platform for company offers, information about
companies and their products. In 2009, the project
was transformed into a price comparator with the
addition of an innovative possibility to request
products besides offering them. The implementation
of the new portal, as described in this paper, was
launched in 2012, and the project will be put into full
operation by the end of 2015 at the latest.
Today, the system is used by users mainly as a
product price comparator. The markets approached
by this project are auction systems, sellers of cars, real
estate, holidays and more.
7 FURTHER SYSTEM
DEVELOPMENT PLAN AND
COOPERATION OPTIONS
The project has a powerful developer base combining
the university environment with the real-world
commercial sphere.
The objective is to become established on the
Central European market (Czech Republic, Slovakia,
Poland, Hungary) with further expansion across
Europe and possibly world-wide.
For these purposes, partners in the regions are
sought in order to make the project viable with a
functional technical and administrative background.
The project has received positive ratings in
several competitions and comparisons; the aim is to
ensure stable project financing by means of subsidies
and community financing.
A precise business model has to be set up, with a
definition of paid system services. With sufficient
financing from public sources, the services could be
offered entirely free of charge for both corporate
clients and the general public.
8 CONCLUSIONS
The paper deals with designing and implementing a
global portal for collecting offers and requests for
specific and special products and services. The portal
works across various sales disciplines and enables
connecting cooperation among products and services.
The paper describes an innovative approach to
entering offers and other connected steps. In addition,
it outlines options for cooperation with other partners
and possible expansion options.
AnIntelligentToolforAll-to-AllSalesontheInternet-PlatformforCompilationofCommercialOffersandRequestsfor
CompaniesandCustomers
269
ACKNOWLEDGEMENTS
This work was partially supported by the Department
of Computer Science, Faculty of Electrical
Engineering and Computer Science in VSB Technical
University of Ostrava, Czech Republic.
This work was partially supported by the Serious
Investment s.r.o. a software development company.
REFERENCES
Ayanso, A., Karimi, A. 2015, The moderating effects of
keyword competition on the determinants of ad position
in sponsored search advertising, Decision Support
Systems, vol. 70, pp. 42-59.
Bauner, C. 2015, Mechanism choice and the buy-it-now
auction: A structural model of competing buyers and
sellers, International Journal of Industrial Organization,
vol. 38, pp. 19-31.
Cai, Y., Mahdian, M., Mehta, A., Waggoner, B. 2013,
Designing markets for daily deals.
Fasuga, R., Drabek, T., Toporkova, G., Holub, L. 2010,
Expert system for data identification in auction system
along with market trends analysis, Proceedings - IEEE
International Conference on E-Business Engineering,
ICEBE 2010, pp. 36.
Fasuga, R., Stoklasa, P., Němec, M. 2014, The method of
automated monitoring of product prices and market
position determination in relation to competition
quotes: Monitoring of product prices and marketability
development with continuous assessment of market
position in on-line sales, ICE-B 2014 - Proceedings of
the 11th International Conference on e-Business, Part
of ICETE 2014 - 11th International Joint Conference on
e-Business and Telecommunications, pp. 5.
Fasuga, R., Toporkova, G., Paluch, M. 2011, Advanced
market trends analysis in electronic auction systems,
Proceedings - UKSim 5th European Modelling
Symposium on Computer Modelling and Simulation,
EMS 2011, pp. 231.
Kannan, A., Talukdar, P.P., Rasiwasia, N., Ke, Q. 2011,
Improving product classification using images,
Proceedings - IEEE International Conference on Data
Mining, ICDM, pp. 310.
Kołaczkowski, P.,Gawrysiak, P. 2011, Extracting product
descriptions from polish e-commerce websites using
classification and clustering.
Kotha, S., Basu, S. 2011, Amazon and eBay: Online
Retailers as Market Makers.
Lahaie, S. 2006, An analysis of alternative slot auction
designs for sponsored search, Proceedings of the ACM
Conference on Electronic Commerce, pp. 218.
Nguyen, H., Fuxman, A., Paparizos, S., Freire, J., Agrawal,
R. 2011, Synthesizing products for online catalogs,
Proceedings of the VLDB Endowment, vol. 4, no. 7, pp.
409-418.
Ong, B. S. 2011, Online shoppers' perceptions and use of
comparison-shopping sites: An exploratory study,
Journal of Promotion Management, vol. 17, no. 2, pp.
207-227.
Overby, E., Forman, C. 2015, The effect of electronic
commerce on geographic purchasing patterns and price
dispersion, Management Science, vol. 61, no. 2, pp.
431-453.
Pham, L., Teich, J., Wallenius, H., Wallenius, J. 2015,
Multi-attribute online reverse auctions: Recent research
trends, European Journal of Operational Research, vol.
242, no. 1, pp. 1-9. Aggarwal, G., Goel, A. & Motwani,
R. 2006, "Truthful auctions for pricing search
keywords", Proceedings of the ACM Conference on
Electronic Commerce, pp. 1.
Wang, C., Zhang, P. 2012, The evolution of social
commerce: The people, management, technology, and
information dimensions, Communications of the
Association for Information Systems, vol. 31, no. 1, pp.
105-127.
Zou, P. J., Cheng, Y., Xu, Y. Y., Fang, Z. 2014, Boost
keywords conversion of search engine.
ICE-B2015-InternationalConferenceone-Business
270