Proceedings Template

Integrated Web 2.0 Style HTTP Referer Analytics

Stephanie Ng

Ateneo de Manila University

Unit 200, 319-H Katipunan Avenue

Loyola Heights, Quezon City

+63-2-9273925

stephanietanng@gmail.com

Ann-Gretchelle Santos

Ateneo de Manila University

58 Unit K Scout Madrinian Street,

South Triangle, Quezon City

+63-2-9267875

gretch.santos@gmail.com

William Emmanuel S. Yu

Novare Technologies – MDI Holdings Inc.

Ateneo de Manila University, Loyola

Heights, Quezon City

+63-2-4266001

wyu@ateneo.edu

ABSTRACT

Little to no attention has been given to referer profiling

which is highly essential in the analysis of monitoring Web site
visits. Web sites involved in e-business are specifically
meticulous about where their visitors come from. This referer
information may provide company Web sites with the adequate
knowledge of visitor interests and what could keep them coming
back. This thesis intended to pay heed to the importance of referer
analysis. It thus created a Web analytics system using Web 2.0
technology with the aid of semantic Web services. Some of the
finest Web services available have been harnessed not only for
data mine precision but also for the expansion of the existing
tools. The developers present an implementation of the work with
the use of the page tagging method in the approach of monitoring
visitor activity in the company Web site. Together with a
scheduled task, this direct method allows for real-time data
processing and the elimination of disk space and bandwidth
compromise on the company's side. Visual reports through the
Web analytics interface are also provided for company viewing.
The usefulness of the work is realized through the company's
need for a sophisticated but extensible tool that will deeper
acquaint them of their visitors.

Keywords

Web 2.0, semantic Web service, referer, Web analytics.

INTRODUCTION

1.1

Background of the Study

With the relentless advancement of Internet technology, the

World Wide Web (WWW) we know today has taken a big leap
from just a mere network of communication for a small group of
engineers to a universal network catering to billions of users
worldwide. It is an avenue for information interchange where
everyone can add to and benefit from the insurmountable data it
holds. Geographical boundaries have also been bridged.
Electronic business, in particular, has laid great advantage to this.
Most companies, regardless of enterprise size, have established
online stores catering to an international audience. eBay and
Amazon have also been instrumental in bringing commodities
closer to the consumer.

Even after the burst of the dot-com bubble, companies are aware
that the WWW is still an indispensable venue for marketing their
products and services. Hence, the concern for Web presence still
prevails. Companies develop Web sites to bridge the gap between
clientele and suppliers. In order to better address the latter's needs,

they harness available Web tools that analyze their users.
However, current tools and technology provide limited user
information. Examples of such are simple counters and user
logging which no longer fulfill the needs of the companies.

Companies are becoming more aggressive in gathering

timely information not only about their visitors but also about
their referers. The referer or the referring page is the URL of the
previous Web page from which a link to the company's Web site
was followed. In other words, the referer specifies the page from
which the visitor accessed the current page. It answers the
question “Where do people come from?” and “What do they like
visiting?”

Referers are important because they signify the community

in which a particular Web site belongs to. As the old adage goes,
“I can know you by knowing your friends.” Referers are friendly
sites. They link and draw visitors to the company's Web site. Not
only are referers important aspects in getting more visitors, they
are also the essential key of getting to know them and what topics
interest them most. This information can help companies devise
the optimal online strategy in designing their Web sites.

However, today's Web statistics focus on user profiling with

little to no emphasis on referer profiling. Referer profiling
primarily deals with categorizing the referer Web sites. Its
variables include title, content, search engine ranking, Web site
keywords, and other pertinent information. The referer area is
often overlooked but as mentioned, it can be the difference
between succeeding and failing online. The developers aim to
supply this necessity in the Web market.

1.1.1

Benefits

1.1.1.1

Significance to Society

This thesis has noteworthy applicability in the domain of E-

business. Companies benefit from receiving valuable insight and
knowledge on the whereabouts and activities of their site visitors.
This Web analytics is able to determine the pages both frequently
and infrequently visited by the company's prospect customers.
Tracking visitors' behavior helps the management devise ways in
increasing usability to ensure that visitors stay engaged and return
to their site.

The Web analyzer not only gathers useful information about

the visitors but also increases the companies' revenue and
conversion. By understanding their visitor's Web usage pattern,
the companies are able to predict an accurate return of investment
(ROI). By monitoring the traffic of customers, the companies may
translate this information into direct sales [1].

Companies are given feedback on how well their Web site is

doing rather than just having it stale online. They are not only
guided to making excellent Web design decisions but also enabled
to “view real data and start using numbers to make educated
marketing decisions [2].” The bottom line is “understanding your
customer is tantamount to making any money in business [2].”

This thesis gives emphasis to referer profiling because it is a

sound and beneficial metric in addressing the common question of
who visits the site. Knowing the referers present linking
opportunities for companies. A mutually beneficial arrangement
such as an affiliate partnership may be created with the top
referers. Both Web sites benefit from the increase in amount of
visitors. For companies that sell advertisements on their Web
sites, this allows them to cater their advertisements to the target
market. If the top referers are Web sites of luxury automobiles,
then it would be wise to place advertisements intended for cars
enthusiasts. Moreover, the Web analytics guides the companies in
making strategic decisions on banner placement.

Through referer profiling, reciprocal linking opportunities

may be explored. Linking has always been fundamental to the
Web. Furthermore, in the past few years, the value of links has
significantly increased. This phenomenon is attributed to the
major search engines, specifically Google's PageRank. “Links
have become the currency of the Web. With this economic value
they also have power, affecting accessibility and knowledge on
the Web [3].” Links now correspond directly to the value, more
commonly known as rank, of a particular site. The more Web
sites that link to a particular site A, the higher the rank of site A
becomes. Not only that, if the linking Web site's rank is
exceptionally high, then it directly increases the rank of site A.

1.1.1.2

Significance to Computer Science

The Web analytics system contributes a novel method of

utilizing semantic Web services in developing the analytics
engine. The use of the semantic Web services is a smarter form of
analysis which goes beyond simple Web counters. Moreover,
these semantic Web services enable the exploration of a new and
in depth strategy of referer profiling. This is an edge because
heavy data processing is done elsewhere rather than having it
done on the server side. This method not only puts existing
services into good use, but it also helps the developers build and
improve on these services.

1.1.2

Scope and Limitation

The analytics tool uses JavaScript for the page tag, as will be

discussed later in the paper. This poses grave problems in cases
when visitors have disenabled JavaScript from their browsers.
This thesis also encompasses the use of existing semantic Web
services but only a small subset of the available resources is
utilized.

The project employs semantic Web services such as Google,

Yahoo!, and MSN search APIs. It also employs software tools
such as phpWhoIS and MaxMind GeoIP (IP Address Location
Technology). However, utilizing semantic Web services brings
about some limitations. The failure or malfunction of the Web
services has a serious repercussion on the functionality of the
Web analytics tool. In addition, the three search engines
mentioned above limit the frequency by which any computer may

access their Web services. If the limit is reached, then the whole
Web analytics is impaired.

It is also important to emphasize that the primary objective

of the project is to develop a system. Whatever patterns derived
from the result is not the focus.

1.2

Project Descriptions

The primary objective of this thesis is to create an integrated

Web analytics using Web 2.0 technology. Multiple best-of-the-
breed semantic Web services, namely Google SOAP Search API,
Google Maps API, Yahoo! Search Web Services and MSN Search
Web Service are utilized to gather timely and accurate
information. Several software tools such as phpWhoIS and
MaxMind GeoIP are also used. The Web application keeps track
of visitor statistics, such as the number of unique visitors, most
popular entry pages, and top referers to a site. The distinctive
quality of this tool is its concentration on referer profiling and
real-time generation of results.

Through the generated results of the analytics, companies

have a concrete measurement of the behavior of visits to their
Web site. They are able to study their visitors - who they are,
where they are from, and how to meet their needs, just to name a
few. Armed with these valuable information, they are able to
increase the usability of their Web site. One specific benefit for
the companies is that they are informed of the landing pages
linked by referers. This, in turn, aids them in making strategic
decisions on banner placement. Companies largely benefit from
boost in traffic, sales, and revenues.

1.3

Research Questions

How can companies monitor the visitor activity of their
Web site in near real-time?

How can companies analyze their visitors without
compromising domain disk space and bandwidth?

How can referers be profiled with the use of integrated
semantic Web services?

METHODOLOGY

2.1

Architectural Design

Semantic Web services such as Google, Yahoo!, and MSN

APIs are utilized in the creation of the analytics. The Web
analytics requests necessary data from these services and
processes the output. Google, Yahoo!, and MSN Search APIs
analyze keywords and return referer page ranks which are used to
profile the Web site's referers. Also utilized are software tools
such as MaxMind GeoIP - which yields the user and referer's
location - and phpWhoIS - which gives detailed information about
the referer's domain.

The collection of the Web analytics data makes use of the

page tagging method rather than the traditional logfile analysis.
Caching presents a problem in logfile analysis. The use of page
tagging, then, ensures that information retrieval is both accurate
and timely.

2.2

Description of Components

Apache

Figure 2.1. Entity Relationship Diagram.

Apache is an open source Web server software that is used to

handle Web requests and serve up Web resources. It runs mostly
on Unix-based operating systems as well as on Windows
platforms. It is the counterpart of Microsoft IIS. Apache is used to
run the developers' PHP pages.

PHP

PHP Hypertext Preprocessor is a Web programming

language that resides on the server. Developers pass data via the
Common Gateway Interface (CGI) from HTML forms for
dynamic content processing which includes database interaction.
PHP scripts are embedded in HTML files. The integrated
analytics project is a Web application therefore PHP is used for
easier Web interface implementation. Specifically, the developers
choose the version 5 because most, if not all, of the Web services
are compatible with the latest PHP version.

MySQL

MySQL is a widely used relational database management

system (RDBMS) that utilizes Structured Query Language (SQL).
It is the most popular open source database because of its
consistent fast performance, high reliability and ease of use [4].
The developers of the Web analytics employ this database for the
storage of valuable client, user, and referer information.

JavaScript

JavaScript is a scripting language interpreted on the client's

Web browser. The script is embedded into a Web page which is
not run on the server side, but on the client side.

2.2.1

Page Tag

The project's method for acquiring the log of a particular

Web page is page tagging. A snippet of JavaScript code, known to
`tag' the visitors, has to be embedded in the client's Web page.

2.2.2

Tracker

An Apache Log Format workalike is utilized to track the hits

of a Web page. The page tag inserted into the client's own Web
page invokes this Javascipt code residing on the server. This code,
called the Tracker, gathers the necessary user and referer
information returned [5]. Examples of such are entry URL, referer
URL, screen resolution, color depth, and visitor IP address.

2.2.3

Logger

The Logger is a short PHP code whose main function is to

store all the information collected by the Tracker into the Job
Queue.

2.2.4

Job Queue

The Job Queue is simply a database where basic information

gathered by the Tracker is stored. These data are accumulated for
later processing.

2.2.5

Scheduled Task

To achieve timely information, data gathering and data

processing are executed independently. Thus, data filed in the Job
Queue are scheduled for processing at a five-minute interval.

2.2.6

Analytics Engine

The Analytics Engine is responsible for pooling collected

data from the Job Queue. It consolidates these data to retrieve
even more information such as title, metatags, and category of the
referer. The detailed records are then entered into the main
database. This is where the referer profiling takes place.

2.2.7

Database

A database of only five tables is used in the implementation.

The tables include: client, logs, visitor, referer, and keyword
information. The client table contains the company's account
information in order for the system to be able to determine which
site to keep track of. The logs table contains the date of access to
the site made by a visitor and the entry URL (landing page) he
visited. The visitor table holds the IP address and the browser
information of the user. The referer table contains the URL, base
url, IP address, title, and category of the referer site. On occasions
when the client's Web site is directly typed and accessed onto the
URL address, the referer data is empty. The keyword table
contains the keywords of each referer obtained through the referer
profiling process. The following is an entity relationship diagram
(ERD) of the developers' database.

2.2.8

Analytics Interface Engine

The Analytics Interface Engine organizes records from the

database into useful statistical information that are
instantaneously displayed on the Analytics Web Interface.
Statistics are obtained through SQL Database Manipulation
Language (DML) or queries made to the database. This solution is
performed in PHP and is responsible for determining information
such as the top entry pages, referers and other pertinent visitor
information from the information system or database.

2.2.9

Analytics Web Interface

This Web interface component is the result of the seamless

combination of multiple semantic Web services and software
tools. Clients are able to register their Web site in order to avail of
the integrated Web analytics. One client is restricted to enrolling
only one URL. They may log on to the Web application to view
the analyzed data of their Web site's user behaviors and activities.
With the provided at-a-glance reports, the clients are able to take
appropriate actions for the improvement of their site performance.

2.2.10

Semantic Web Services

The developers use the following semantic Web services in

the retrieval of the categories and other pertinent information of
client's referers. Google, Yahoo!, and MSN searches provide
different sets of data. Integrating them generates more reliable
results. The developers aim to ensure that the information
retrieved is accurate. If one search engine fails to generate a result
set, the two others may compensate.

2.2.10.1

Google SOAP Search API

The Google API is a free experimental Web tool that

developers can use to find and manage information on the Web
\cite{google}. It is a service in which Google offers their
resources to developers and researchers for the latter's own
application development. The software applications created by the
developers remotely connect to the service through an XML-
based system, SOAP.

2.2.10.2

Google Maps API

The Google Maps JavaScript API lets developers embed

Google Maps in their own Web pages. These maps are
customizable to the needs of the users. Markers and lines can be
drawn on the map to specify certain locations.

2.2.10.3

Yahoo! Search Web Services

Like Google API, Yahoo! Search Web Services allows

access to Yahoo content and services. However, unlike Google,
Yahoo! uses REST instead of SOAP. Yahoo! claims that it has a
lower barrier to entry, easier to use, and entirely sufficient for
their services in contrast to SOAP. Yahoo! Web Service is also
language independent which gives developers freedom and ease
in integrating the service into their application [7].

Under the umbrella of Yahoo! Search Web Services, three

data sources are particularly utilized. First is the Web Search
service that allows developers to search the Internet for Web
pages. Second is the Term Extraction service which provides a list
of significant words or phrases extracted from a larger content. It
sifts out all the irrelevant words and returns the keywords for the
content passed. It uses the very algorithm Yahoo! Search engine
uses to rank pages. Third is the Site Explorer service which
provides access to information on individual sites. The InLink
data service is a subset of this and gives Web site owners a good
grasp of who is linking to their site. It retrieves the information
about the pages linking to a particular Web page.

2.2.10.4

MSN Search Web Service

MSN Search Web Service, like Google, also uses SOAP. It

also allows users access to some of MSN's services. It also
allocates a separate search query quota to each IP address
allowing third-party users to use developed applications at a more
extensive level [8].

2.2.11

Software Tools

The developers also employ several open source software

tools to provide clients more details regarding their Web sites.
Given the IP address, MaxMind GeoIP determines geographical
location and phpWhoIS retrieves the domain owner information.

2.2.11.1

MaxMind GeoIP (IP Address Location

Technology)

Geographic location of a website or website visitor is

identified real-time by tracking the user's Internet Protocol
address. GeoIP determines the country, region, city, postal code,
and area code of the visitor as well as provide information such as
longitude/latitude, connection speed, ISP, company name, domain
name, and whether the IP address is an anonymous proxy or a
satellite provider [9].

A new “Know Your Customer” law has been implemented

by regulatory entities in both US and Europe. As a result, banks,
software vendors and other online enterprises are subject to
compliance, thus providing a more secure environment.
Geolocation is used as an investigatory tool by the security teams
to track Internet routes of online assailant and prevent future
assault from the same location. Hosts of live video streaming such
as internet movie vendors and online broadcasters are able to
monitor their viewers on licensing regulation conformance. This
technology is very functional in several industries, including e-
retail, banking, media, online gaming and law enforcement, for
preventing online fraud, complying with regulations, and
managing digital rights. It also provides location-based content
such as the user's language, currency, and pricing.

GeoIP obtains its dataset from the user's area or zip code

information entered upon the filling out an online form. The data
is then run through a series of algorithms that acquires sets of IP
addresses in the particular location. The next time the site is
visited by the user's neighbor, GeoIP already guesses his/her
location.

This is utilized in identifying the referer's location,

specifically country, city, longitude and latitude, in this project's
attempt to be an integrated analytics.

2.2.11.2

phpWhoIS

phpWhoIS is a software tool that identifies the owner of a

domain given its URL. This is used for additional information
gathering.

2.3

Implementation Details

Clients first need to embed a JavaScript code, called a page

tag, into every page of their Web site that they wish to track. The
page tag invokes a tracker, another JavaScript code responsible
for gathering basic referer data such as referer URL, entry URL,
browser resolution, and color depth. The tracker then passes the
referer data to the logger, a PHP code, of which its only task is to
store these data into a Job Queue for later processing. Data in the
Job Queue does not take long, however, since the records are
processed every five minutes. A component relationship diagram
is shown in Figure 2.2.

The analytics engine processes the information from the Job

Queue. It also acquires more information about the referring Web
site. It gets the meta tags of the referer Web page and parses the
actual contents to obtain its title and body.

The engine also executes the first phase of the analysis

process which is the referer profiling. Referer profiling simply
means categorizing the referer Web sites. However, this is the
most complex part of the implementation process. Keywords are
the vital determinants of categories. Three different sources are
mined to resolve the keywords: meta tags - specifically the
keywords and descriptions, the title, and the body itself. The
combined text of the sources is passed to the Yahoo! Term
Extraction Web Service which returns the list of significant words
and phrases in order of importance. The top ten words and phrases
are the keywords. Each one is run through the three search
engines. The search engines will return the position or rank of the
referer Web site based on the keyword used. The site may be any
from the first through the 100th search result displayed. Google,
Yahoo, and MSN all rank the pages differently. The developers
seek to ensure accuracy which is why all three are utilized. The
numeric ranks are added to get the total page rank for the
particular keyword used. After all the keywords have been
processed, the total page ranks are compared. The least, that is,
the most popular, keyword will be utilized as the category.

All the mined information are then entered into the database.

This real-time process guarantees accuracy of the results. Once
the data gathering and profiling is done, the database is populated.

The Analytics Interface Engine executes the second phase of

the analysis process. It simply evaluates information previously
filed in the database by organizing them for display. It
accomplishes this task simply by SQL queries.

The clients can log in to the Analytics Web Interface or the

Web application to view their Web site statistics. From this point,
Web Services and software tools continue to process and return
data. The geographical locator gathers the country, city,
longitudinal and latitudinal information from the IP addresses
stored. Google Maps API gives a visual perspective to these
geographical locations. Special emphasis is given to the referers,
their entry pages, and possible banner placement.

Everything is developed using PHP scripts understood from

Google, Yahoo!, and MSN APIs and other tutorials found online.

Figure 2.2. Component Relationship.

RESULTS AND DISCUSSION

3.1

Near Real-Time Monitoring of Visitor

Activity

The approach for companies in monitoring the visitor

activity of their Web site is through page tagging. The task is as
simple as inserting a snippet of JavaScript code in every page of
the Web site that the company wishes to submit for Web analysis.
The code then invokes another JavaScript code, the tracker with
the company's user ID as its request parameter for identification.
The company's Web page containing the page tag must be
rendered for the page tag to execute and invoke the tracker script
to capture data. The tracker gathers basic referer data which is
passed to a PHP code, the logger. The logger is responsible for
storing the data amassed into a job queue.

Although data in the job queue is stored for later processing,

it does not take long since the records are processed every five
minutes. Five minutes is near real-time. A shorter time frame will
cause too much overhead and overlapping in the database.

The following is a sample of the page tag script (Figure 3.1)

that is returned to the user upon registering for an analytics
account. Following it is the scheduled task (Figure 3.2),
specifically the cron daemon, that configures the job queue into
processing every five minutes.

The page tagging that the developers use is specifically

designed for near real-time processing. In addition, not all

information that may be acquired from Web services are stored,
but are rather retrieved when the user views the reports to ensure
the timeliness of data.

Figure 3.1. Sample Page Tag Script with Company ID 37

Figure 3.2. Scheduled Task Script Used to Trigger

Job Queue Every Five Minutes

3.2

No Compromise in Domain Disk Space

and Bandwidth

In contrast to a hybrid scheme of collecting Web analytics

data through combined logfile analysis and page tagging,
dedicated page tagging is employed in the implementation of this
thesis. It does not make use of the conventional 1x1 image
requested by the page tag thus creating an entry in the client's
logfile. Moreover, Web server logs are not designed to mine data
but simply to debug Web servers. They are flat files on multiple
file systems and even stored in different time zones. It is even

more complicated if Web servers are distributed geographically in
different time zones, especially for large sites that have multiple
servers - each logging data into separate files and on different file
systems - since combined data requires it to be in one time zone
(usually GMT). Web server logs also store large amounts of
structured data inefficiently because they are typically in ASCII.
Redundant and irrelevant information may also be contained in
Web server logs. This is evident in every request for images
found in a page. Using server logfiles for analyzing visitor
activities is therefore inefficient [10]. Logfile analysis also poses
inaccuracies due to browser caching. If a visitor revisits a page,
the second request is oftentimes retrieved from the browser's
cache so that request is received by the Web server only once.
This traditional method does not only deposit large amounts of
data into the client's logfile and waste disk space but it also leaves
the data for later analysis.

Dedicated page tagging allows companies to gain reports of

visitor profiles without compromising their domain disk space and
bandwidth since there is neither client nor server-side processing
involved. Rather, computation and analysis are remotely done on
the developer's technology and stored in a dedicated domain thus
saving the user from expensive space that can be better used in
storing their own data. The job queue, in particular, makes the
task possible because background processing is done every five
minutes. As a result, the clients outsource the analytics system
and by doing so, they free up their own system.

Another advantage for the analytics to be located external to

the company's domain is that the reports need not be restricted to
the viewing of the Web site developer. The analytics caters its
organized reports to the owner, manager or any marketing contact
of the company. These reports are statistics derived from SQL
queries made to the database.

3.3

Referer Profiling with the use of

Integrated Semantic Web Services

Referer profiling simply means categorizing the referer Web

sites. Keywords are the vital determinants of categories. In order
to establish the profiling procedure, the analytics engine obtains
the necessary information about the referer. The information
consists of the meta tags, the Web site title, and the body itself.
These three are integrated in order to produce a collection of texts
to be passed to the Yahoo! Term Extraction Web Service. The
Term Extraction Web Service returns the list of keywords
arranged in order of importance. To limit the scope and to
expedite the processing function, only the top ten keywords are
considered. Each keyword is run through Google, Yahoo, and
MSN search Web services. They return the rank of the referer
Web site based on the particular keyword searched. Since the
three search engines all have different algorithms and distinct
results, all are utilized to ensure accuracy. The sum of the three
ranks is the total rank for that keyword. After all keywords have
undergone the same procedure, the keyword with the highest rank
is deemed as the category.

Using best-of-the-breed semantic Web services is a novel

idea which places heavy data processing particularly on Google,
Yahoo, and MSN. It is quite advantageous because the
developer's server becomes even more dedicated to the Web
analytics itself. It is able to give more processing power to
gathering information and producing the client Web site reports.

CONCLUSION

In this thesis, the developers have discussed the importance

and benefits of Web analytics. The analytics built by the
developers has a particular emphasis on referers. Analyzing
referers is an advantage in knowing a visitor's background of
interest, thus gaining knowledge on what types of visitors
frequent a Web site. The construction of the analytics has also
been possible with Web 2.0 technology. Harnessing an array of
semantic Web services has not only allowed a more precise data
yield but has also extended the development of the services. This
method has also kept heavy data from being processed on the
client or on the server's end. Although the software produced uses
a variety of semantic Web services, the implementation is not a
Web service itself, but rather, a consumer of it. In principle, the
accomplishment is part of the Web service eco-system, if termed
as such. The task has also been to create a tool - although not any
ordinary tool - the referer Web analytics tool.

Companies can monitor the visitor activity of their Web site

in near real-time. It has been conferred that data is processed and
stored near real-time thus preserving the integrity of the
information. The use of the technology, in consequence, will give
users access to timely referer profiles. This will thus inform
company users of strategic banner placement, for example.
Companies can also analyze their visitors without compromising
their domain disk space and bandwidth. There is neither client nor
server-side processing involved because clients outsource the
analytics system and by doing so, they free up their own system.
These have been possible through the page tagging method and
the scheduled task. Referers can be profiled with the use of
integrated semantic Web services. The three search engines
employed in the development all perform different algorithms and
distinct results, all of which are used to ensure accuracy.

The developers have undoubtedly presented a sophisticated,

yet extensible Web referer analytics with the stated features. It is
without reservation that a tool such as the current implementation
is vastly essential in e-business today.

RECOMMENDATIONS

The work has been an initial attempt at referer profiling in

Web analytics. Although the technology is sufficient in its right,
the developers propose additions to the work to further extend its
possibilities. This includes adding wider ranges of Web services
and accordingly tracking more information. Security issues must
also be resolved such as those involved in the page tag. The user
id must be encrypted, for example, and the logger script
information must not be disclosed.

ACKNOWLEDGMENTS

We thank God who has gone with us in the creation up to the
completion of this thesis. We thank Him for giving us wisdom
and strength. We also thank the Department of Information
Systems and Computer Science of the Ateneo de Manila
University and our thesis adviser Mr. William Emmanuel S. Yu
for their constant support during the entire process of this thesis.

REFERENCES

[1] THINKMETRICS. Thinkmetrics web analytics - providers

of website analysis and statistics. Available from
http://www.thinkmetrics.com.

Document Outline