Link-US

Document Sample
Link-US Powered By Docstoc
					    Link-US
“scheduling made easy”
      A Data / Web Mining Project



         As a Requirement for:

     MIS510: Data and Web Mining



      November 27, 2011



     Project Group
         James Chen
            Chad Ho
     Jonathan Shacter
Link-US – “scheduling made easy”



Table of Contents


TABLE OF CONTENTS                                  2

TABLE OF FIGURES                                   3

INTRODUCTION                                       4

1. SYSTEM ARCHITECTURE                             6

1.1 APIs                                            7
   1.1.1 Google Map                                 7
   1.1.2 Yahoo! Local                               7
   1.1.3 SMSEverywhere.com                          7
   1.1.4 WeatherForYou.com                          8

1.2 Data Mining Tools                               8
   1.2.1 Weka                                       8

1.3 Other Tools                                     8
   1.3.1 XML/SWF Charts                             8
   1.3.2 Http Client                                9

1.4 User Interface                                  9
   1.4.1 OSWD Template                              9
   1.4.2 Dreamweaver 8.0                            9
   1.4.3 Adobe Photoshop CS2                        9


2. BUSINESS MODEL                                  10

2.1 Advertising                                    10
   2.1.1 Google AdSense                            10
   2.1.2 Target Businesses                         10

2.2 Subscription                                   11

2.3 SMS Text Messages                              11


3. NOVELTY                                         12

3.1 Similar Services in the current market         12
   3.1.1 Directory Websites                        12
   3.1.2 Experience-Sharing Websites               12

3.2 How Link-US Differs                            12
   3.2.1 Scheduling                                12
   3.2.2 Business Model                            13
   3.2.3 Technology                                13


                                             -2-
Link-US – “scheduling made easy”

4. FUTURE WORK                           14

4.1 APIs                                 14
   4.1.1 Routing                         14
   4.1.2 Weather Forecast                14

4.2 Data Mining                          14


CONCLUSION                               16

REFERENCES                               17



Table of Figures
FIGURE 1 - SYSTEM ARCHITECTURE            6




                                   -3-
Link-US – “scheduling made easy”



Introduction

Link-US is a combination of the words “link” and “us”. “Link” represents the idea of linking users to
various sites, and “us” stands for both “us” in the context of a group and “US” as in the United States.
The site is designed to ease the schedule planning process within the United States; therefore, the name
“Link-US” is most appropriate.


Many websites exist which allow users to search for restaurant or entertainment information and some
even offer recommendations; however, the goal behind Link-US was to take this planning functionality to
the next step by offering a scheduling feature which allows user’s to create and share events with friends,
as well as, a recommendation system based off users’ schedules in order to provide places where users
will most likely go.


The system allows registered users to enter their personal preferences for both restaurants and
entertainment, and then uses these preferences to provide recommendations. User’s can also utilize our
visualization in order to help see what other people are doing and what places are popular in particular
cities.


The goal of Link-US is simple; we want to make the schedule planning process faster, easier, and more
reliable.


Before a user would have to search for places online, gather the information, filter out places based off
personal preferences, perform some sort of mining on the data (usually done in a person’s head), and then
try to make an executive decision based off what was collected. There are two major problems with this:
time and the lack of all potential data.


Link-US does the filtering process for the user and mines the data in order to get the best potential results
for a user. The user can make use of visualizations rather than text-based data in order to make better
decisions. The best part about this is it reduces the time needed to plan an event.


The site can be found at the following URL: http://128.196.239.75:8080/Link-US/index.jsp




                                                    -4-
Link-US – “scheduling made easy”

The remainder of this report will discuss the details of the architecture, business model, novelty, and
future work to consider.




                                                    -5-
Link-US – “scheduling made easy”



1. System Architecture

In order to accelerate our development cycle and avoid any unexpected bugs, we decided to apply the
most mature technologies. Since the focus of this project was on data and web mining, the use of
multiple services was necessary in order to fetch data or provide visualization for analysis.


Therefore, Java becomes the best option to us.




                                       Figure 1 - System Architecture


The following is a brief overview of the system architecture:


    -   The Tomcat Server and JSP Dynamic Content pages were chosen as the technology to implement
        our web interface.
    -   Our customers will access these JSP pages to access our contents and services.



                                                    -6-
Link-US – “scheduling made easy”

    -   We store our data into a MySQL database because it is one of the best free open source database
        solutions and it offers high performance which helps fulfill our data mining need.
    -   Weka was used to make the association rules based on schedule history. Our system matches the
        user’s queries with the most relevant results and provides recommended options to the user based
        on the association rules created in Weka.
    -   In order to get the local business content based on entertainment or restaurants, we decided not to
        build our own spider, but to leverage the current most completed web contents provided by
        Yahoo Local.
    -   Furthermore, we include SMS text messaging feature, weather feeds, and Google maps for user
        convenience.


Specific details of the architecture are presented as follows:



1.1 APIs


1.1.1 Google Map

The maps feature is one of the key components in our system. We used the Google Map API because it
provided a very convenient way to access its service. Basically, we just need to use JSP to generate
necessary Javascript content to access this service. This API also provides us the function to pop-up an
information window to display business data.



1.1.2 Yahoo! Local

Yahoo Local is a service which provides local business information to Internet users. It provides a set
API to help users retrieve business information, including title, address, phone, website URL, and ratings.
This is our main data source for website contents which are further recorded into schedule tables for data
mining activities.



1.1.3 SMSEverywhere.com




                                                     -7-
Link-US – “scheduling made easy”

A lot of time was spent on searching different SMS APIs, because we wanted our customers to have the
ability to send event invitation through SMS text messaging to any of their friends. We decided to use the
service from SMSEverywhere.com, because it provided the most flexibility and had no associated cost
involved.



1.1.4 WeatherForYou.com

We decided to provide weather information to our customers because it is very critical for our customers
when they are trying to pick up an activity based on weather. We did a research for most of famous
weather information provider on the Internet. However, we found that most of the APIs are too
complicated and the implementation could be very time consuming. So, we chose to access the service
from weatherforyou.com. It provides simple yet comprehensive weather information, including graphics,
for cities in the United States.



1.2 Data Mining Tools


1.2.1 Weka

Weka is a tool used to generate the association rule for our project. We use it to store the results of the
association rule into our database for our recommendation system. It only functions in batch mode so we
have to update it manually in order to recalculate the association rules.



1.3 Other Tools


1.3.1 XML/SWF Charts

Charts are important to our web design for visualizing the valuable information to our customers. Based
on a classmate’s recommendation, we tried this component and found it was very easy to implement.
This tool provides many types of charts, but we chose to use the column, bar, and polar charts to best
represent the data.




                                                     -8-
Link-US – “scheduling made easy”


1.3.2 Http Client

In order to automatically retrieve third party data and web services in the background, we used Http
Client as a replacement of browser to interact with third party web services. This component also can
help us to access some services on the websites, which don’t wrap their web content into web services.



1.4 User Interface


1.4.1 OSWD Template

A template for the site design was found at oswd.org. The template was designed by Wink Hosting and
tailored to for our site. Background, colors, and text were all changed in order to create a nice site layout.



1.4.2 Dreamweaver 8.0

Dreamweaver 8.0 was used to develop the JSP (Java Server Pages) dynamic content pages used on the
site. This type of page was chosen because it was what the group was most familiar with. We needed
dynamic content pages in order to link our MySQL database to a more user friendly page.



1.4.3 Adobe Photoshop CS2

Adobe Photoshop CS2 was used to edit and create some of the images found on the site. Images added to
the overall appeal and visually helped the site.




                                                     -9-
Link-US – “scheduling made easy”



2. Business Model

Link-US is dependent on three types of revenue streams: advertising, subscription, and SMS text
message.



2.1 Advertising

There are a few different ways to go about advertising: Google AdSense, target businesses based on our
customer data, or a combine the two.



2.1.1 Google AdSense

Google AdSense allows site owners to place ads on their sites and receive compensation from Google
when they are clicked. This will allow Link-US to have Google display relevant ads (entertainment and
restaurants) on our site, and will compensate us when they are clicked by our users. This stream of
revenue is small since it is dependant on users clicking the links; however, it will bring in a small
percentage of operating capital.



2.1.2 Target Businesses

The Link-US customer base help provide useful information such as popular restaurants or entertainment
in a city based on scheduled events for that city. If used correctly, this information can be extremely
useful to local business when they come up with their marketing plans. Businesses can realize customers’
preferences, consuming behaviors, and compare the two of them to further determine how best to market
their business. For example, Tucson’s Yamato Japanese Restaurant may discover from our site that
people who prefer Japanese food go to Sushihmaha more frequently. Using this information they can
analyze why it happens and determine a course of action such as distributing promotional material or
advertisements to our users with a Japanese food preference. Once our customer base is large enough,
Link-US will have a powerful channel.




                                                    - 10 -
Link-US – “scheduling made easy”

This channel can be sold at a premium or simply used to target businesses and offer them advertising
space at a cost yet to be determined.



2.2 Subscription

We plan to charge a yearly subscription fee of $25 per user to enjoy the features of our site. This is just
over $2 a month to enjoy custom schedules, easy to use features, recommendations, and the other great
options Link-US offers.



2.3 SMS Text Messages

We plan to charge customers $0.02 per SMS text message. This can potentially be the major source of
our income. Assuming a customer sends 1000 total messages a year, that is $20 that customer will pay in
SMS costs. Now figure we have 1000 users; that is $20,000 in operating income just from text
messaging. By charging a tiny fee per message, Link-US is able to bring in a large amount of operating
income.




                                                   - 11 -
Link-US – “scheduling made easy”



3. Novelty

3.1 Similar Services in the current market

There are two major types of restaurant and entertainment websites: director and experience sharing
websites.



3.1.1 Directory Websites

Directory websites provide the listing function of the business information. The most famous directory
websites is yellowpage.com. Basically, yellowpage.com applies only advertisement model.



3.1.2 Experience-Sharing Websites

Experience-sharing websites provide the discussion function for users to make evaluations and comments.
The most famous experience-sharing website is Zagat survey. It started as a local restaurant magazine in
New York City, and now has become the most popular online restaurant and entertainment information
provider. It requires users to subscribe in order to access information and give comments and feedback.



3.2 How Link-US Differs

Link-US differs from the above mentioned sites by having the following novelties:



3.2.1 Scheduling

Scheduling is our major novelty because the original two types of websites are built to only feed user’s
information regarding restaurants and entertainment. However, users need more than just a listing or
searching tool, and should be given a means to save their search results. Most people who use a
restaurant or entertainment information website are looking for places they can go with friends or family,
not just by themselves.



                                                  - 12 -
Link-US – “scheduling made easy”



Possibly the most frustrating thing about the other sites is that after spending an entire afternoon
searching for a place to eat or something to do, it takes the rest of the night to create a schedule and make
phone calls to invite all the people you want.


In order to reduce the user’s time and frustration, scheduling has become the key point to make Link-US
superior than other restaurant/entertainment information websites. Link-US leverages the current well-
built business information from Yahoo Local and provides custom recommendations from our own
database of what other users are doing. This helps eliminates the filtering process done by the user
before, and reduces the overall time it takes to find a place that meets a user’s preferences.



3.2.2 Business Model

The business model itself has already been explained, but we this is also part of our novelty. Link-US
will apply an advertisement model and subscription model at the same time. The accumulated scheduling
data collected will also be a valuable asset to Link-US. We can do further mining on the user’s behavior
based on time or geographic (city/state), and then sell this information to businesses. Based on our
analysis, we will be able to clearly identify the most potential registered members and further offer the
targeted marketing service to our business customers by sending the promotional information by SMS or
email.



3.2.3 Technology

The most important technological novelty in Link-US is the combination of a recommendation system
and scheduling system. In addition, SMS service is another technological novelty to let users send out
their schedules at any time and from any place. Finally, Link-US provides easy to use, comprehensive,
chart tools to help users analyze data and generate better schedules based off the information provided.




                                                    - 13 -
Link-US – “scheduling made easy”



4. Future Work

Although Link-US has already provided many useful and friendly functions, such as SMS invitation
service and customized recommendation for businesses based off personal preferences, our team believes
that there are still some other functions which can make Link-US more powerful.



4.1 APIs


4.1.1 Routing

Currently, the Google Map API is used to display the maps which allow users to easily and clearly know
where the locations of each event are. Google Map API, however, does not provide direction function. It
is inconvenient for users to get directions between many locations. Therefore, Link-US will improve the
routing function for schedules. When users go to the printer friendly page, they can also directly print the
directions without typing locations many times at Google Map.


The potential solution could be to use the “MapQuest API”. It integrates MapQuest's ultra-accurate
driving directions which can show our users exactly how to get the destination.



4.1.2 Weather Forecast

Weather is a very critical consideration for travelers when they schedule a trip. Automatically providing
weather forecast data for trip planners will be a sweet design. In addition, our event recommendations
could be based on weather forecast. If it will be a rainy day tomorrow, for example, Link-US will alert
users and recommend some indoor activities so that our user can always have fun time.



4.2 Data Mining

Our event recommendation is based on the association rule derived from other user’s schedule data. In
the future, we would like to categorize our users based on their preferences by some unsupervised




                                                   - 14 -
Link-US – “scheduling made easy”

machine learning algorithm. By mining the schedule data of users who are in the same category, Link-US
can provide more accurate and more relative recommendations.




                                                - 15 -
Link-US – “scheduling made easy”



Conclusion

Link-US is an extremely easy to use system that provides a lot of functionality for the user. The
scheduling system is unique to any restaurant / entertainment information website, and is the next step in
online event planning.


Easy to use visualizations provide a unique way of looking at text-based data and allow the user to make
better decisions. The SMS text messaging feature is a cool concept that is making use of popular
technology.


The business model is sound and targeted ads will be easy to once the systems users grow.




                                                  - 16 -
Link-US – “scheduling made easy”



References

1. Referenced 2007, from http://www.google.com/apis/maps/


2. Referenced 2007, from http://developer.yahoo.com/search/local/V3/localSearch.html


3. Referenced 2007, from
   http://smseverywhere.com/send.htm#SENDERhttp://smseverywhere.com/send.htm#SENDER


4. Referenced 2007, from http://www.weatherforyou.com/


5. Referenced 2007, from http://www.maani.us/xml_charts/


6. Referenced 2007, from http://www.oswd.org/


7. Referenced 2007, from http://www.zagat.com


8. Referenced 2007, from http://purplepages.com


9. Referenced 2007, from
   https://www.google.com/adsense/login/en_US/?hl=en_US&sourceid=aso&subid=ww-en_US-et-
   ads1_D&medium=link




                                                - 17 -

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:8
posted:11/27/2011
language:English
pages:17