Christopher C. Hartwell
Parker H. Sternbergh
Tulane University
 

Abstract

This project examines the problem of search engine accessibility in terms of what is commonly called "the invisible web." As the problem pertains to community resource web sites, many are presently available on the internet with content that is not indexed by the major search engines. This makes the resources more difficult for consumers to find.

This project examines current and past research regarding search engine accessibility and search engine optimization. It goes on to demonstrate a method for making invisible web data (in the form of community resources) accessible to the major search engines. It is a project designed to aid the community service organization professional in implementing their organization's web site in order to better serve all stakeholders. The outcome of our project is a web site that provides community resource data to the public through the major search engines, and we present limited qualitative data which demonstrates its effectiveness.




Acknowledgements

The authors would like to acknowledge the major search providers Google and Yahoo! for providing their search services to the public free of charge. We also acknowledge the many contributions of authors in the field of search engine optimization and search engine marketing, who over the years have helped to guide the ethical (or "white hat") practice of search engine optimization. These contributors are not limited to Danny Sullivan (Founder and Editor of Searchenginewatch.com), Jill Whalen (Owner, Highrankings.com), Brett Tabke (Owner, Webmasterworld.com) as well as the many members who post in the forums of Webmasterworld.com. We would also like to acknowledge the Louisiana Office for Addictive Disorders for putting together the database of treatment facilities on their web site, and Tulane University School of Social Work for accepting this topic as a Capstone Project.

Problem Formulation

The growth of the internet brings many opportunities for non-profits and community service organizations to reach their clients. When organizations take on the task of implementing a web site, decisions must be made about the budget for the site, the graphical design of the site, which information to include on it, and how to present the information to the public. Furthermore, significant thought must be given to where financial resources should be spent. It is necessary to determine what should be spent on the actual design of the site, as well as what should be spent on marketing the site.

There are many ways to make community resources available to the public through web sites. One way is to raise awareness of a web site through advertising. Organizations commonly place ads on television, billboards, promotional products, and radio in order to raise awareness of themselves and the services they provide. This form of advertising can bring direct traffic to a web site, because the public may remember the web address and type it directly into a browser. Another way to make these resources available to the public is through advertising on other web sites. Ads placed on web sites are usually hyperlinked to the site that the ad is promoting, bringing the user directly to the site when the ad is clicked. While both methods are commonly used by organizations to raise web site visibility or awareness, search engines have frequently been overlooked because they are poorly understood, even by many web site designers. In our discussion, we will sometimes mention terms that traditionally aren't a part of the educational background received by professionals who are employed in community service organizations. In order to facilitate better understanding of our project in this field, we have included an appendix with definitions that correspond to the words in bold print within this text.

In order to gain some understanding of how complex searching the web has become, one must consider the increases in usage. In 1996, there were 7.5 million searches on the web. (Brin, Page, 1998). By 2000 the number increased to 41 million (Davis, 2002), and according to Nielsen//Netratings, in January 2005, there were 4,086,000,000 searches on the web. Between 1994 and 1997, the average number of documents available to a web searcher increased 20 fold, and today, on April 16, 2005, Google reports an index of 8,058,044,651 documents that it searches (Brin, Page, 1998; Google, 2005). In February of 2004, Nielsen//Netratings reported that 39 percent of Americans used a search engine during January 2004, and that the 114.5 million unique users, represented 76 percent of the active online U.S. population.

The increasing usage and size of the web has started many new industries. There are specialists who study huge compilations of web search query logs in order to understand aggregate user web search behavior and query patterns. The objective is to design sites that better anticipate how a user searches for data and to incorporate that knowledge into site design and search engine optimization (SEO) strategies. Silverstein and Henzinger (2002) conducted one of the largest of such studies. The subject of the study was the Alta Vista Search Engine. The group studied a query log with one billion search request entries and 285 million user sessions over a six week period (Silverstein, Henzinger, 2002). Silverstein and Henzinger (2002) discovered that "web users differ significantly from the user assumed in the standard information retrieval literature." The group found out that users in the log typed short phrases (instead of word strings connected by "AND" and "OR" for example). In another study, Joachim (2002) determined that there was an average of 2.35 words in the average query. Users typically looked at the first ten items in the retrieval list (Silverstein, Henzinger, 2002). In fact, 85% looked only at the first screen of a query and very few queries were revised (Silverstein, Henzinger, 2002). Healthcare has been a primary search topic on the web. By 2002 healthcare sites were among the fastest growing page categories on the web (Davis, 2002). In a statistically significant poll conducted in 2001, 40% of respondents reported using the internet to get advice or info about health or healthcare (Baker, Laurence, Wagner, Todd, Singer, Sara, Burndorf, Kate 2003).

In addition to the increasing size of the web and its usage, hardware and crawling technology must keep up with the staggering growth. Search engines have multiplied and become very competitive. According to Nielsen//Netratings, as of January 2005, Google had 47% of the market followed by Yahoo at 21%. Search engines may be publicly accessible or private and each has its own policy and method of crawling the Web. Each search engine also has varying technical ability to keep up with the size and volume of traffic on the web.

Search engines do not index sites equally (Brin, Page, 1998; Lawrence, Giles, 2000). Some of the best academic literature on this subject has come out of Stanford University and the founders of Google. The founders of the most popular search engine, Google, created the search engine while at Stanford University. Their main focus was a method of determining the importance of a web page in relation to the user's query. The method that evolved as a determination of importance of a web page is still in use today, represented by a factor that Google calls PageRank. PageRank responded to the challenges of web growth, the increase in queries and indexed pages, and the problem of finding meaningful responses to queries. PageRank has been widely studied and has become a respected standard in the industry. Because the principles of PageRank have been applied amongst the major search engines, it was important that we study it and incorporate its principles into our web site design methodology. In Page's definitive 1998 paper discussing the rapid growth of the World Wide Web, they did a random sample of servers to investigate the amount and distribution of information on the Web. They pointed out that in there were approximately 800,000 million pages and over 3 million servers. At that time, the largest search engine only covered 30% of the web (Brin, Page, 1998).

The internet contains a mix of viable academic and reliable resources as well as less credible commercial information and resources. Brin and Page (1998) determined at the time of their study, that 83% of search engines had commercial content. They further determined that about 6% of servers have scientific and, or educational content. At the time it was found that overlap between search engine results was low (Brin, Page, 1998). In their study, Brin and Page (1998) found that it took 57 days to get a page registered in a search engine, the only way the site would be available for indexing. Today, this is not the case, modern day search engines index sites rapidly once they are found through crawling hyperlinks.

For the scope of this project we examined current knowledge of search engine accessibility and search engine optimization. Many community service organizations, from private therapists, to government agencies have web sites; and many people are now using search engines to find these web sites (Sullivan, 2005). The problem which occurs is that community resource web sites are not always designed with search engine accessibility and, or optimization in mind. When this happens, potential clients are not able to find the information that they are looking for. Often times, data is contained in an online database such as the community resource data for Southeastern Louisiana accessible through the Via Link web site (http://www.vialink.org) and the Louisiana Office for Addictive Disorders (LOAD) database of treatment facilities (http://www.dhh.state.la.us/offices/locations.asp?ID=23). Databases such as these contain a large amount of useful information including names and descriptions of local agencies; but if users don't already know the organization's web site address or happen upon the site through a hyperlink from another site, the data never gets found. This phenomenon is commonly known as "the invisible web."

The invisible web consists of online databases, excluded web pages, and many types of web pages that are actively generated by the server upon the user's request (as opposed to "static" html documents) (UC Berkley, 2004). Many private networks (usually password protected) and private databases are also considered a part of the invisible web, but for our purposes we will focus on the content that is intended to be available to the public. This invisible web exists because of present day limitations to what search engines add to their index. A search engine's "index" is its own database of web pages that it draws from in order to produce search results. It finds data by "spidering" or "crawling" web pages from hyperlink to hyperlink. Many invisible web databases are accessible only through their respective database search forms; therefore the search engine spider has no way of accessing the data (Raghavan, Garcia-Molina, 2000; Lawrence, Giles, 2000).

The goal of this project is to illustrate a method of making an "invisible web" database more accessible to search engines. The need for solutions to the community resource gap in the search engines is evidenced by many web sites that provide community resources online, yet do not make these resources accessible to search engines. For the purpose of our project, we have used the Louisiana Office of Addictive Disorders' database of Louisiana's substance abuse clinics, treatment centers, and prevention programs as an example of an invisible web database. The database uses query strings in the URL's of the resources available, and does not provide a way for search engine spiders to index the resources. When web sites are designed without search engine accessibility in mind, this data is harder to get to because one needs to know exactly where to go in order to access the data - which in this case is the following web address: http://www.dhh.state.la.us/offices/locations.asp?ID=23.

The web site we created in order to demonstrate our solution to the invisible web problem is "Louisiana Addiction Resources" (LAR) and it can be found at http://louisiana-addiction.com. The objective of this project is to provide awareness of search engine accessibility issues, a method of solving the problem, and guidelines for community service organizations to implement an optimized and accessible web site.

Methodology

Our project was done as a descriptive design with limited qualitative analysis. As evidence that accessibility works, we will present qualitative data that internet users typed into search engines. The data presented are queries which resulted in those users finding our web site.

In this section we will discuss common themes related to search engine accessibility and optimization that we used to construct our community resource web site. Our first task was to become familiar with the subjects of "the invisible web" and "search engine optimization." The invisible web exists because documents and data aren't accessible to the search engine for a number of possible reasons, and the first step in creating our web site was to outline a static hyperlinked structure that the search engine spiders could follow. Here is an outline of the link structure we used:

I. Home Page

a. Louisiana Parishes Page (with links to a page for every parish in the state).

i. Individual Parish Pages (with links to a page for each parish's resources)

1. Resource 1

2. Resource 2

3. Resource 3

4. Etc.

While accessibility is the first factor in making internet resources available through the search engines, without optimization, a site can still fail in its purpose. The concept of optimization is a sub-topic of accessibility that is almost just as important; because without optimization, frequently, even accessible documents aren't able to be found by the searcher. We will discuss optimization further in the next section.

Search engine spiders do not presently crawl many types of "dynamic" web pages for many different reasons (dynamic refers to types of web pages which are created by the server upon the user's request) (UC Berkley, 2004; Raghavan, Garcia-Molina, 2000; Lawrence, Giles, 2000). Dynamic web pages are often a product of database driven web sites; and since most community resource data available through government web sites and online directories is taken from databases, much of this content is not crawled by search engines. As mentioned earlier, two examples of dynamic databases that are presently not indexed by search engines are the Louisiana Office of Addictive Disorders (LOAD) database (http://www.dhh.state.la.us/offices/locations.asp?ID=23) and the Via Link database (http://www.vialink.org/index1.html).

The solution we propose is to have static hyperlinks that lead to a directory of the database's content. In order to accomplish this for the LAR web site, we created a taxonomy for organizing the resources in directory format. Examples of similar solutions, on a much larger scale, are DMOZ.org (The Open Directory Project), Google Directory and Yahoo! Directory. All are static, crawl-able, taxonomies of web sites covering almost every topic available on the internet. The LAR web site we have created in order to illustrate this technique, is an all inclusive directory of the same data that is contained in the LOAD database. The information contained within LAR is in the format of static html pages, linked together by static hyperlinks, which are easily crawled by present day search engine technology, making the information more widely accessible to the public.

Database query strings and session ID's are one of the most common barriers to accessibility for search engine spiders when attempting to index dynamic content. Search engine spiders are instructed not to follow these types of hyperlinks because of a multitude of different problems that can occur (UC Berkley, 2004). Here are two examples of URL's with database query strings (the first from the Via Link web site, and the second from the LA Office of Addictive Disorders web site):

http://unity.servicept.com/cp/findhelp/findhelpbasic.php?sid=5095761fc45efe1fb945d37a67ef85bd&rand=1113213131

http://www.dhh.state.la.us/offices/locations.asp?ID=23&Detail=151

Database query strings and session ID's are characterized by question marks (?), and equal signs (=) in the URL, often having "?ID=" in that particular order. A typical static web page URL will look similar to this:

http://louisiana-addiction.com/parishes/lincoln-parish.shtml

There are more characteristics to consider when optimizing URL's, and we will discuss those in the next section.

Optimization Guidelines

Next, we had to become familiar with the important concepts and methods involved in designing well-optimized web sites. The guiding principle of search engine optimization is to design a web site that shows up in the search engines for queries related to its content. Not only does a well-optimized page tell the search engine what the page is about, but it also has links from other pages on the internet that indirectly serve to tell the search engine what that page is about. With that being said, we can divide the concept of optimization into two distinct processes, "Off-Page Optimization" and "On-Page Optimization." Recent advances in search engine technology have taken into account historical data in their method of ranking search results as well (Anurag, Cutts, Dean, Haahr, Henzinger, Hoelzle, Lawrence, et. al. 2005). Although worth mentioning, we will only go into brief detail regarding this factor since it is so new and so little is known about it, and also because it can get more technical than is appropriate for this paper.

On-Page Optimization

Google maintains much more information about web documents than typical search engines. Every hitlist includes position, font, and capitalization information. Additionally, we factor in hits from anchor text and the PageRank of the document. Combining all of this information into a rank is difficult. We designed our ranking function so that no particular factor can have too much influence (Brin, Page 1998, p.1).

Proximity of words on a page also plays an important factor in the ranking of web pages in search results. For example, a search for the phrase "marriage counseling" will (with all other factors being equal) pull up a page that has "marriage counseling" in its text before a page in the results that has "marriage and family counseling" in its text. The proximity of words plays an important role in the search results as explained below.

For a multi-word search, the situation is more complicated. Now multiple hit lists must be scanned through at once so that hits occurring close together in a document are weighted higher than hits occurring far apart. The hits from the multiple hit lists are matched up so those nearby hits are matched together. For every matched set of hits, a proximity is computed. The proximity is based on how far apart the hits are in the document (or anchor) but is classified into 10 different value "bins" ranging from a phrase match to "not even close" (Brin, Page 1998, p.1).

URL's play a role in search engine optimization as well. While some URL's can make a page inaccessible, others can help to optimize a page.

Pages created as the result of a search are called ‘dynamically generated' pages. The answer to your query is encased in a web page designed to carry the answer and sent to your computer. Often the page is not stored anywhere afterward, because its unique content (the answer to your specific query) is probably not of use to many other people. It's easier for the database to regenerate the page when needed than to keep it around (UC Berkley, 2004, p.1).

As evident by the result, the Louisiana Addiction site has 253 web pages indexed. In the Yahoo! search engine, one can simply type in the URL without the "site:" command to perform the same test.

For this project we focused on the area of addiction resources in the state of Louisiana. We found that the state web site (Louisiana Office if Addictive Disorders or "LOAD") was a perfect example of an online database that was not accessible to the search engines.

The next step in our project was to begin a web site for Louisiana Addiction Resources that would be accessible to the search engines. In order to start the web site, we chose the name "Louisiana Addiction Resources" and registered the URL "Louisiana-addiction.com".

The web site was designed from the ground up, with search engine accessibility in mind. Parts of the URL for all pages were done descriptively with page and directory names relevant to their content. The same was done for the Title tags of the HTML pages as well. Where words were included as image files in the web pages (such as the logo), alternate text was used in the HTML; and the title of the site (Louisiana Addiction Resources) was also written in plain text within a header tag on the homepage. All hyperlinks, title tags, and header tags were done descriptively as well, with hyperlinks in plain text.

Next, we added our resources to the web site. The agencies were also added to the site with search engine accessibility in mind. Each agency included in the site had a web page devoted to it with the name of the agency in text on the page as well as in the title of the page.

Off-Page Optimization

PageRank is a concept that was developed by Google founders Sergey Brin and Lawrence Page when they developed the Google search engine. "PageRank is an excellent way to prioritize the results of web keyword searches. For most popular subjects, a simple text matching search that is restricted to web page titles performs admirably when PageRank prioritizes the results (demo available at google.stanford.edu) (Brin, Page 1998)." The PageRank concept is still in use on Google today, and is symbolized in their directory and on their toolbar by a measurement of 0 - 10. Google uses a graphic image of a green bar over a white bar to show the PageRank, or "importance" of a web site in their directory.

PageRank is currently updated almost quarterly for all web pages, and "invisible web" or inaccessible pages never receive a PageRank score. Using the Google toolbar (http://toolbar.google.com) is an unscientific way of indicating whether or not a URL has been indexed. If the PageRank indicator is all white or grey in color, it could mean that a site is either brand new, has PageRank "0," is penalized, or not yet included in Google's index. (Web pages are sometimes penalized when they are found to be using techniques such as hidden text, doorway pages, and deceptive redirects, aimed at deceiving a search engine). A PageRank score can be from 0 - 10, with "10" being one of the most important pages on the web, and "0" being one of the least important.

Google determines PageRank using many factors. Although search engines don't make their algorithms public for obvious reasons, there is much we know from what they have made public, past research, and experience in the field. According to Google's web site:

PageRank relies on the uniquely democratic nature of the web by using its vast link structure as an indicator of an individual page's value. In essence, Google interprets a link from page A to page B as a vote, by page A, for page B. But, Google looks at more than the sheer volume of votes, or links a page receives; it also analyzes the page that casts the vote. Votes cast by pages that are themselves "important" weigh more heavily and help to make other pages "important" (Google, 2004).

<a href="http://www.example.com">Anchor Text</a>

Creates a hyperlink that looks like this:

Anchor Text

and is linked to the web site http://www.example.com.

Some words from Google's founders on anchor text:

The text of links is treated in a special way in our search engine. Most search engines associate the text of a link with the page that the link is on. In addition, we associate it with the page the link points to. This has several advantages. First, anchors often provide more accurate descriptions of web pages than the pages themselves. Second, anchors may exist for documents which cannot be indexed by a text-based search engine, such as images, programs, and databases (Brin, Page 1998, p.1).

IBM researchers as well have noted the importance of anchor text in the indexing and ranking of documents:

Anchor text is typically very short, and provides a summarization of the target document within the context of the source document being viewed. Our main premise is that, on a statistical basis at least, anchor text behaves very much like real user queries. For this reason, a better understanding of the relationship between anchor text and their target documents will likely lead to more effective results for a majority of user queries (Eiron, McCurley 2002, p.1).

Homepage (filename: index.shtml)
Treatment Facilities (text that links to treatment-faclilities.shtml)

Louisiana Parish Treatment Facilities (filename: treatment-faclilities.shtml)
Acadia Parish (text that links to acadia-parish.shtml)
Allen Parish (text that links to allen-parish.shtml)
Ascension Parish (text that links to ascension-parish.shtml)

As one can see on the homepage (filename: index.shtml), the anchor text "Treatment Facilities" is used to link to the page with the parishes listed. This was done with the intention of helping the page to show up for searches like "Acadia Parish Treatment Facilities." It tells the search engine that this page has to do with "treatment facilities." Ideally, the link should tell the search engine that this page is about the Louisiana Parishes and their treatment facilities, but usability must come first; in other words, "treatment facilities" was chosen so that the site's visitors will know where to go from the homepage to find their resources (and because "Louisiana Parishes Treatment Facilities" is too long and cumbersome for the text of a main menu item). It is a blend of search engine optimization and usability.
Here is an example of how the parish pages are linked to their individual resources:

Each link on the page has descriptive anchor text, and brings the user to the document it describes (the facility's database record). Quite frequently, web sites use images to anchor hyperlinks to other documents. Those images sometimes represent pictures and icons, but frequently have words on them as well. The problem associated with using image files to anchor hyperlinks, is that they frequently give no descriptive information to the search engine spider because the web designer forgot to use alternative text in the html document. "Alternative text" or "alt" text is contained within the "alt" attribute of the image html tag, and it is used simply, to describe the image file. Here is an example of the html that would be used to insert a well-optimized image file (a picture of two dogs playing in a yard):

<img src="dogs-playing.jpg" alt="Dogs Playing in a Yard">

"Alt" tags are used for other purposes as well, but only in the case of images contained within hyperlinks are they presently indexed and used by Google.

Anchor text should also be considered when asking other web sites to link to your site. This applies for internet directory listings as well. Since we wanted our web site to be found for the search "Louisiana Addiction Resources," we asked for that link text in our directory listings rather than making the title of the listing "LAR," as we refer to it in this context. If we want the site to show up for both "LAR" and "Louisiana Addiction Resources," it's beneficial to vary the link text occasionally. For a good list of web directories that can help your site gain link popularity, visit http://www.strongestlinks.com/directories.php.

Also worth mentioning are meta-tags. Meta tags are a part of the html head section where an author can define the keywords they would like associated with their web page, as well as a description. While there was a time when the use of meta-tags was popular and had an affect on search engine rankings, today their affect is minimal if anything at all. The problem was that too many authors abused the use of meta-tags in order to get their pages to rank well for irrelevant keywords. In response, search engines stopped using meta-tags to affect rankings. Some still include meta "descriptions" in their search results.

Historical Data

Historical data is a factor that has recently evolved into the ranking of web sites in search results. In a recent Google patent application they discuss many factors such as the age of links pointing to a site and the length of time domain names are registered (Anurag, Cutts, et. al. 2005). Many links that appear all at once pointing to a site can be seen as an attempt to deceive a search engine through buying links from other web sites. The document explains an analysis of a site's natural progression of link building and rewards sites that adhere to that progression (Anurag, et. al., 2005). This information is a likely explanation for some phenomenon noticed by search engine optimizers over the past year which they have labeled "the sanbox." The sandbox theory describes the fact that many new web sites have not been able to rank well for competitive keyword phrases, even though the sites were well-optimized, and had good links pointing to them. Some describe the sanbox as lasting from 6 mos. to 1 year in length. While this information is useful and relevant to any new web site, it is included here as a concept to be aware of when designing a site; we do not wish to go into too much detail here, as it would stray too far from the intended scope of the project. If the agency's need is to target a competitive market that already has many well optimized sites in the search results, it should hire an experienced SEO professional to do the work.

Competitiveness of Search Terms

Not all search terms are created equal! Some phrases are more common than others, therefore statistically; a web page has less of a chance of showing up on the first page of results when the topic it covers is something very common and on millions of other web pages.

This competitiveness of search terms also applies to phrases that are commercial in nature. Wherever there is money to be made on the web, there are web site owners trying to show up in search results. This leads to a competitiveness that may exist regardless of the commonality of the word or phrase. Commercial competitiveness can hinder a social service oriented web site's ability to rank well for the search terms they target. One advantage that social service agencies have, however, is that they are more likely to be able to get other web sites to link to their site free of charge due to the "helpful" nature of their services.

Participants

The participants in the project are Christopher C. Hartwell, Parker H. Sternbergh, the LOAD web site, as well as all substance abuse clinics, treatment centers, and prevention programs that were included in the site. Other stakeholders are internet users who seek the information the site provides through search engines, either directly, or those who happen upon the site by chance as well as internet users in general, who all have an opportunity to happen upon this web site.

Data Analysis

The final method used to show the effectiveness of our project was to collect information from the server logs contained on the computer that hosts the LAR web site. In particular we extracted query strings that were typed into search engines by users, which resulted in that user finding the data contained within our web site. The LAR site (at the time of this project) was hosted on a Unix server by Aplus.net of San Diego, California. The statistical program used to extract the data from the server logs was Mach5 Analyzer version 4.1.5 created by Mach5 Enterprises, LLC. The data shown here come from server logs that span a 31 day period from March 16, 2005 to April 16, 2005. The LAR web site has been online in an uncompleted form starting on November 5, 2004. All of the agency data was added between February and March of 2005.

The following table is a 31 day list of keyword phrases that were typed into search engines which resulted in that internet search engine user entering the LAR web site. The number to the right represents the number of times each phrase was used, and the search engine listed is the referring site.

Click Here for Search String Data

The above search strings are all associated with text contained within the LAR web site.

To compare the site's traffic before and after it was crawled by the search engines produces a clear and predictable result since the site was promoted by no other means than the DMOZ.org link. Prior to being indexed, the site had an average of 1 visitor per day (in November 2004), and after being indexed it maintained a more steady stream of traffic, which has increased since November 5, 2004 to where it is now at 43 visitors per day. The following chart depicts the site's traffic from the beginning to where it stands today. "Visits" are the number of unique users that visit the site.
Summary by Month

Summary by Month
Month	Daily Avg				Monthly Totals
	Hits	Files	Pages	Visits	Sites	KBytes	Visits	Pages	Files	Hits
Apr 2005	211	165	123	43	451	22265	824	2346	3147	4025
Mar 2005	161	121	110	32	496	25889	992	3417	3780	4993
Feb 2005	150	56	35	13	174	14942	374	1006	1589	4206
Jan 2005	15	3	9	5	82	659	172	297	110	465
Dec 2004	24	9	9	5	79	1214	169	293	295	765
Nov 2004	71	17	5	1	19	1619	39	138	452	1848
Totals						66588	2570	7497	9373	16302

Although the web site is near completion and it has been indexed by all of the major search engines, we have limited qualitative data to analyze. We do know to some degree, that our project has been successful in its goal, because it does show up for many search queries related to the site's content. (Such as "Louisiana Addiction Resources" and queries involving parish names, treatment facilities, etc.) The outcome of the project demonstrates a method of building community resource web sites which are accessible to search engines, and therefore, more accessible to the public. The practical implications of this project are that community resource organizations and government agencies who follow the guidelines described herein, will be able to make their web sites more accessible to the public.

The broad implications of this project are largely unknown at this point. We would like to assume that, because community resource data is more accessible, that consumers are finding the resources and putting them to use. However, this can not be known without further study. Entrance and, or exit polls on the web site would be potentially useful for finding out more information about the internet users who visit the site. Potential information worth gathering would be:

• Is this visitor a person seeking resources for themselves, or is he or she a healthcare professional who is looking for resources in order to help a client?
• Did the visitor find the resources they were looking for through our web site?
• If the visitor did not find exactly what they were looking for, did they find something comparable?
• Did the user actually follow through with a referral from our web site (by contacting one of the agencies listed)?
• If a referral was followed through with, was it for the visitor, or for another individual?
• Which visitors to the web site are more likely to follow through with referrals (consumers or healthcare professionals)?
• Which visitors to the web site are more likely to follow through with referrals (visitors who find the site through search engines, by word of mouth, through links from other web sites, or other means)?

Many new questions are raised by this research, and the creators of this project believe that the internet provides great opportunity for the future of information and referral services for community service organizations and the healthcare profession.

Limitations

The solution we have employed to search engine accessibility may not be the most efficient solution for every web site. Our solution requires updating the database separately from the web directory. There are however, many other ways to make invisible web databases more accessible, depending on the database technology in use. For some web servers, what's called a mod_rewrite can be done in order to make the URL's created by the server appear to be static to the search engine (Whalen, 2004). As Jill Whalen (2004) points out in her newsletter, mod_rewrite may not always be necessary as search engines are increasingly getting better at indexing dynamic content. What is currently necessary is that there is some form of static hyperlink to the content, dynamic or not. The use of the scripting language PHP (Hypertext Preprocessor) has become a very efficient means for interfacing a database with html content, and can be done in ways that are invisible to the search engine. PHP is quickly becoming the search engine friendly method of choice for implementation of database driven web sites among search engine optimizers.

A limitation to the visibility of the LAR web site in search results is the fact that it is relatively new, and no money has been spent promoting the site. The stats obtained for this project are a result of the site being initially found and indexed by the search engines through a free DMOZ.org directory listing. Were the site promoted more through online directory listings and links from other web sites, it would have ranked higher for more competitive search terms.

Another limitation of this project was the fact that our analyses of results were limited due to the fact that we only have statistics from the LAR web site. Even if we could get stats from the LOAD web site, there would be too many extraneous variables to account for the traffic the site receives to make a real comparison. Our original project sought cooperation with Via Link to make some of their data accessible, a project that could have been measured using before and after comparisons; however, Via Link never followed through with providing us with the sample data. Since the LAR site has been accessible from the start, we don't have a good opportunity to make a before and after comparison with its statistics.

The information contained herein is as current and up-to-date as we could provide. Search engine technology changes at a rapid pace in order to keep up with the ever-expanding world wide web, and in order to provide searchers with the most relevant results. Time sensitivity of research data is an important consideration when implementing a web site, because what was important one or two years ago (like meta-tags), can be near irrelevance today.

Discussion

We live in a time when many people begin their search for services on the internet (Baker, et. al., 2003). Searching the internet is a private and easy way to begin to collect information and resources needed to help solve many problems. Consumers may not be aware that an invisible web exists; therefore, an agency that provides community services should take care to be listed on a web page that is accessible to search engines and avoids the problem of becoming a part of the invisible web. Invisible web databases, such as LOAD discussed earlier, are useful to the public who begin their search or find their way to the LOAD web site; however, search engine accessible databases provide enhanced efficiency in making resources available to the public by allowing them to be included in their index as well. As organizations become more aware of search engine accessibility as an issue, newer and better solutions will be implemented; and ideally, more community resources will fall into the hands of the people who need them.







References

Anurag, Cutts, Dean, Haahr, Henzinger, Hoelzle, Lawrence, et. al. (2005). Information retrieval based on historical data. U.S. Patent and Trademark Office, Patent Application Full Text and Image Database. Retrieved April 16, 2005 from http://appft1.uspto.gov/netacgi/nph-Parser?
Sect1=PTO2&Sect2=HITOFF&p=1&u=%2Fnetahtml%2FPTO%2Fsear
ch-bool.html&r=1&f=G&l=50&co1=AND&d=PG01&s1=2005007
1741&OS=20050071741&RS=20050071741.

Baker, L., Wagner, T., Singer, S., Burndorf, K. (2003). Journal of the American

Medical Association. 289 (18): 2400-2406.


Brin, S., Page L. (1998). The Anatomy of a Large-Scale Hypertextual Web Search

Engine. Computer Science Department, Stanford University, Stanford, CA USA.


Davis, J. ( 2002). Disenfanchising the Disabled: The Inaccesibility of Internet-Based Health

Information. Journal of Health Communication. 7: 355-361.


Eoron, N., McCurley, K. S., (2003). Analysis of Anchor Text for Web Search. IBM Almaden Research Center. Retrieved April 16, 2005 from
http://www.almaden.ibm.com/cs/people/mccurley/pdfs/anchor.pdf.

Google (2004). Our Search: Google Technology. Retrieved April 16, 2005 from http://www.google.com/technology/.

Google (2005). Google. Retrieved April 16, 2005 from http://www.google.com.

Joachims, T. (2002). Optimizing search engines using clickthrough data. Proceedings of the ACM Conference on Knowledge Discovery. USA, 2002

Lawrence, S., Giles, C. L., (2000). Accessibility of information on the Web. Association of Computing Machinery v.11, n.1, pp. 32-39.

Nielsen//Netratings (2004). One in Three Americans Use a Search Engine, According to Nielsen//Netratings. Retrieved April 16, 2005 from http://www.nielsen-netratings.com/pr/pr_040223_us.pdf

Nielsen//Netratings (2005). Majority of Online Searchers Use Multiple Search Engines, Pointing to a Fluid Competitive Landscape, According to Nielsen//Netratings. Retrieved April 16, 2005 from http://www.netratings.com/pr/pr_050228.pdf

Raghavan, S., Garcia-Molina, H. (2000) Crawling the Hidden Web. Computer Science Dept., Stanford University. Retrieved April 16, 2005 from http://dbpubs.stanford.edu:8090/pub/2000-36.

Silverrstein, C., Henzinger, M. (2002). Analysis of a Very Large Web Search

Engine Query Log. Palo Alto: Google.com.


Sullivan, D. (2002) How to Use HTML Meta Tags. SearchEnginewatch Dec. 5, 2002 Retrieved April 16, 2005 from http://searchenginewatch.com/webmasters/article.php/2167931
Sullivan, D. (2005). Neilson NetRatings Search Engine Ratings, SearchEngineWatch.,

January 2005. Retrieved April 12, 2005 from

http://searchenginewatch.com/webmasters/article.php/2167931.



UC Berkley Library (2004). Invisible Web: What it is, Why it exists, How to find it, and Its inherent ambiguity. Regents of the University of California. Retrieved April 16, 2005 from http://www.lib.berkeley.edu/TeachingLib/Guides/Internet/InivisibleWeb.html

Whalen, J. (2004). High Rankings Advisor Issue 102: Preparing for Your SEO Campaign - To Mod_Rewrite or Not. High Rankings Advisor - Issue No. 102 Retrieved April 16, 2005 from http://www.highrankings.com/issue102.htm






Bibliography

Anurag, Cutts, Dean, Haahr, Henzinger, Hoelzle, Lawrence, et. al. (2005). Information

retrieval based on historical data. U.S. Patent and Trademark Office, Patent

Application. Full Text and Image Database. Retrieved April 16, 2005 from

http://appft1.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p
=1&u=%2Fnetahtml%2FPTO%2Fsearch-bool.html&r=1&f=G&l=50&co1=AND&d=PG01&
s1=20050071741&OS=20050071741&RS=20050071741


Baker,,Laurence,Wagner, Todd,,Singer,Sara,Burndorf,Kate. (2003). Journal of the

American Medical Association. 289 (18): 2400-2406.


Banach, Mary,Bernat,Frances. Liability and the Internet : Risks and

Recommendations for Social Work Practice. (2000). Journal of Technology in

Human Services. 17 (2/3) : 153-171.


Barratt, Moira. (2003). Organizational Support for evidence-based practice within child

and family social work: a collaborative study. Child and Family Social Work: 8

(2): 143-151.


Baker,Laurence,Wagner, Todd,Singer,Sara,Burndorf,Kate. (2003). Journal of the

American Medical Association. 289 (18): 2400-2406.


Brin, S., Page, L. (1998). The Anatomy of a Large-Scale Hypertextual Web Search

Engine. Computer Science Department, Stanford University, Stanford, CA USA.


Cravens, J. (2000). Virtual volunteering: online volunteers providing assistance to

human service agencies. (2000). Journal of Technology in Human Services.

17(2/3): 119-136.

.
Davis, Joel. ( 2002). Disenfranchising the Disabled: The Inaccessibility of Internet-Based

Health Information. Journal of Health Communication. 7: 355-361.


Eoron, N., McCurley, K. S., (2003). Analysis of Anchor Text for Web Search. IBM Almaden Research Center. Retrieved April 16, 2005 from http://www.almaden.ibm.com/cs/people/mccurley/pdfs/anchor.pdf.

Gardyn, R. (2002). Surfing For Health. American Demographics. 3: 14-15.


Gilfords, E.D. (1998). Social Work on the Internet: an introduction. Social Work. 43

(3): 243-251.


Google (2005). Google. Retrieved April 16, 2005 from http://www.google.com.

Google (2004). Our Search: Google Technology. Retrieved April 16, 2005 from http://www.google.com/technology/.


Heinlan, K., Welfel, E., Richmond, E., Rak, C. (2003).

The Scope of Web Counseling: A Survey of Services and Compliance With

NBCC Standards for WebCounseling. Journal of Counseling and Development.

81(1): 61-69.



Holden, G., Finn, J. (2000). Human Services Online: A New Arena for Service

Delivery. Part II. Journal of Technology in Human Services. !7 (2/3): 295-297.


Joachims, T. (2002). Optimizing search engines using clickthrough data. Proceedings of the ACM Conference on Knowledge Discovery. USA, 2002

Kibirige, H., Depalo, L. (2000). Information Technology and Libraries. 19: 1-13.


Lawrence, S., Giles, C. L., (2000). Accessibility of information on the Web. Association of Computing Machinery v.11, n.1, pp. 32-39.

Mandl, K., Felt, S., Pena, B., Kohane, I. (2002). Archives of

Pediatrics and Adolescent Medicine. 154(5): 508-511.


Marks, Jerry D. (2000). Online Fundraising in the Human Services. Journal of

Technology in Human Services, l1( 2/3) 137-152


McCarty, D., Clancy, C. (2002). Telehealth : implications for social work practice.

Social Work. 47(2):153-161.


Meir, A. (2000). Offering Social Support via the Internet: A case Study of an Online

Support Group for Social Workers. 17(2/3): 237-266.


Levine, J. (2000). INTERNET: a framework for analyzing online human service

practices. Journal of Technology in Human Services. 17 (2/3): 173-192.


Nielsen//Netratings (2005). Majority of Online Searchers Use Multiple Search Engines, Pointing to a Fluid Competitive Landscape, According to Nielsen//Netratings. Retrieved April 16, 2005 from http://www.netratings.com/pr/pr_050228.pdf

Nielsen//Netratings (2004). One in Three Americans Use a Search Engine, According to Nielsen//Netratings. Retrieved April 16, 2005 from http://www.nielsen-netratings.com/pr/pr_040223_us.pdf

Patterson, D. (1996). An electronic social work knowledge base: a strategy for

global information sharing. International Social Work. 39:149-61.


Powell, T., Jones, D., Cutts, D. (1998). Web site engineering:

beyond Web page design. New Jersey: Prentice Hall.


Price, G., Sherman, C. (2001). The Invisible Web: Uncovering Information

Sources Search Engines Cant See. Internet: Cyberage Books.


Raghavan, S., Garcia-Molina, H., (2000) Crawling the Hidden Web. Computer Science

Dept., Stanford University. Retrieved April 16, 2005 from

http://dbpubs.stanford.edu:8090/pub/2000-36


Richie, H., Blanck, P. (2003). The promise of the Internet for disability: a

study of on-line services and web site accessibility at Centers for Independent

Living. Behavioral Science & the Law. 21(1): 5-23.


Safran, Charles (2003). The collaborative edge: patient empowerment for vulnerable

population. International Journal of Medical Informatics. 69(2/3): 185-191.


Schultz, J.A., Frncisco, S.B., Wolff,T, et al. (2000) The Community Tool Box:

Using the Internet to Support the Work of Community Health and Development.

Journal of Technology in Human Services. 17(2/3):267-293.


Silverrstein, C., Henzinger, M. (2002). Analysis of a Very Large Web Search

Engine Query Log. Palo Alto: Google.com.


Sullivan, D. (2002) How to Use HTML Meta Tags. SearchEnginewatch Dec. 5, 2002 Retrieved April 16, 2005 from http://searchenginewatch.com/webmasters/article.php/2167931

Sullivan, D. (2005). Neilson NetRatings Search Engine Ratings,

SearchEngineWatch., January 2005. Retrieved April 12th,2005 from

http://searchenginewatch.com/webmasters/article.php/2167931.


Sullivan, D. (2005). Share of Searches: April 2004. SearchEnginewatch Retrieved

April 12th,2005 from http://searchenginewatch.com/reports/ article.php/2156451


Sullivan, D. (2005). Share of Searches: December 2004. SearchEngineWatch.

Retreived April 12th from http://searchenginewatch.com/reports/ article.php/2156451.

UC Berkeley Library (2004). Invisible Web: What it is, Why it exists, How to find it, and its inherent ambiguity. Regents of the University of California. Retrieved April 16, 2005 from http://www.lib.berkeley.edu/TeachingLib/Guides/Internet/InivisibleWeb.html

UC Berkeley Library (2004). Meta-Search Engines. Regents of the University of California. Retrieved April 16, 2005 from
http://www.lib.berkeley.edu/TeachingLib/Guides/Internet/Meta-Search.html


WebWorkshop (2004). Google's PageRank Explained and How to Make the Most of It.

Retrieved from http://www.webworkshop.net/pagerank_calculator.php.


WebWorkshop (2004). Search Engine Optimization (SEO) - the basics (how to improve

your search engine rankings. Retrieved March 15, 2005 from

http://www.webworkshop.net/pagerank_calculator.php.


Wyatt, Jeremy. (1997). Commentary: measuring quality and impact of the world wide

web. Information in Practice. British Medical Journal. 314: 1879.




Appendix

Anchor Text- The text that is used to link from one web document to another.

Black Hat Search Engine Optimization- Search engine optimization strategies that attempt to deceive or mislead the search engine.

Crawler or Spider- A program that automatically retrieves web pages. Spiders are used to feed pages to search engines. It's called a spider because it crawls over the web from hyperlink to hyperlink.

Database- A collection of information organized in such a way that a computer program can quickly select desired pieces of data.. An electronic filing system. Traditional databases are organized by fields, records and files. A field is a single piece of information; a record is one complete set of fields; and a file is a collection of records.

Dynamic Pages- Web pages that are actively generated by the server upon the user's request. Most often they are generated in response to database searches. Dynamic pages are the opposite of static web pages, which are documents that reside in their full form on the server at all times.

HTML- Short for Hypertext Markup Languages used to create documents on the World Wide Web..HTML defines the structure and layout of a web document by using a variety of tags and attributes.

Proximity- The degree of separation of words within a document.

Hyperlink- An element in an electronic document that links to another place in the same document or to an entirely different document. Hyperlinks bring the user to the linked document by way of a mouse click.

Invisible web- that part of the web that cannot be indexed by search engines. One of the most common reasons that a Web site's content is not indexed is because of the site's use of dynamic databases. Dynamic databases can trap a spider. Web pages can also fall into the invisible Web if there are no links leading to them, since search engine spiders typically crawl through links that lead them from one destination to another.

Off-Page optimization- Factors that affect a document's search engine ranking from outside of that document.

On-Page optimization- Factors that affect a document's search engine ranking from within the document.

Meta tag- A part of the head section of an HTML document that provides information about a web page. Unlike normal HTML tags, meta tags do not affect how the page is displayed. Instead, they provide information such as who created the page, how often it is updated, what the page is about, and which keywords represent the page's content.

PageRank- Google's proprietary method for ranking web pages. A measure of "importance" of a web page.

Private Network- A network that has limited access to it's pages.

Query Strings- Query strings come into being when a user types in a search term(s). At that point the search engine or database driven web site will create a dynamic URL based upon the query. Query strings typically contain ? and % characters. They can sometimes be a barrier to search engine spiders.

Search Engine- A program that searches documents for specified keywords and returns a list of the documents that, ideally, best match the user's query. Examples of search engines are Google and Alta Vista. Typically, a search engine works by sending out a spider to retrieve as many documents as possible. These documents are then indexed based on their content.

Search Engine Optimization- the process of helping a document to show up in search engine results for queries related to its content.

Session ID- The identifier for a specific session completed by a specific user. Session ID's are contained within the URL's of dynamic web sites and can sometimes be a barrier to search engine spiders.

Static Page- A page that is fixed and not capable of action or change. A web site that is static can only supply information that is written into the HTML and this information will not change unless the change is written into the source code. When a web browser requests the specific static web page, a server returns the page to the browser and the user only gets whatever information is contained in the HTML code. In contrast, a dynamic web page contains content that a user can interact with, such as information that is tied to a database.


URL- Short for Uniform Resource Locator, the global address of documents and other resources on the World Wide Web. The first part of the address indicates what protocol to use, and the second part specifies the IP address or the domain name where the resource is located. Typically appearing like "http://www.example.com."

Visible Web- The part of the Web that a search engine is able to access with a web crawler.

White Hat Search Engine Optimization- Search engine optimization strategies which attempt to influence a search engine to rank a document high in the search results for queries related to its content. "White Hat" optimization techniques do not attempt to deceive the search engine.