Business Sales Expert System

1
Business Sales Expert System
2
Intro

• As the Internet becomes more accessible, it is
  important to build more sophisticated systems on
  the web.
• These sophisticated web systems are often
  referred to as lightweight systems.
• Web systems generally use dhtml, xml, javascript,
  php, coldfusion and other scripting languages.

3
Business Sales

• The goal of this project was to create a rule
  based system for sales people. The system is
  meant to help sales people show customers more
  relevant items, hoping to expedite the sale.
• We used a Jewelry Point of Sales System as our
  starting point. This helped restrict our domain
  space.

4
Asynchronous Javascript and XML (AJAX)

• For this project we used AJAX to develop the
  system and create a robust interface for making
  the recommendations.
• AJAX is a unique combination of Javascript and
  XML, allowing web pages to be dynamically changed
  without refreshing the page.
• Example http//local.google.com

5
Step 1

• First we must select a customer from the
  database.
• For this project, I was able to get a sanitized
  version of the Jewelry Systems database. The
  database contains sales history, but minimal
  customer information.
• To Sanitize the database we shortened all the
  last names to 3 characters long and removed all
  personal information.

6
Step 2

• Our rules are based on the customers sales
  history and what the customer has been
  recommended before.
• After selecting the customer we look at the
  customers sales history and select the types of
  Jewelry items they usually shop for. For
  example, if they usually shop for bracelets and
  necklaces, we will recommend items in those
  categories.
• In this step we extract the category types, metal
  types and price range the customer usually
  purchases within.

7
Step 3

• We then take a Cross Product of Category Types
  and Metal Types, giving us a wider range of items
  to show the customer.
• We also make sure to show the customer things
  they havent seen in the last 3 months. 3 months
  was chosen based on the average frequency of
  customers to this Jewelry store.

8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
Not Interested

• If the customer is not interested in the items
  shown, all they need to do is click on the not
  interested button and new items will appear.
• These items populate the page following the same
  price, category and metal type restrictions as
  the beginning.
• At this point, if there are not more items within
  the specific search, the price range is widened
  by 25 on each end.
• This will continue to update until there are no
  more items to show.

12
(No Transcript)
13
Conclusion

• It has become clear to me that web systems are
  going to become a large force in the future of
  computer systems. Therefore, It is important to
  incorporate Artificial Intelligent techniques
  into them as early as possible. I hope to
  continue working with this system and create more
  rules to make the system an Expert.
• The site is hosted at www.heart32.com/718Project
• username 718
• Password project