On May 1, 2001, a beta test of IconFind's IICS implementation plan and corresponding coding architecture was initiated. Clear objectives were laid out and controlled procedures put in place in order to provide objective data for evaluation. For purposes of increased relevancy, this test was conducted over the Internet in real-time where actualities and not simulations could all be taken into account and placed in the body of empirical evidence.
The areas we decided to measure and study in this beta test were simple and straightforward though far reaching in effect. They are as follows:
1. Using the iics categorization code outside the framework of programming or markup languages such as HTML or XML - By making it possible to place the categorization code on the Web pages without regard to the strict requirements of the various programming and markup languages, the entire process is greatly simplified making it available to virtually everyone.
2. Length of time it takes search engines to index iics categorization code on categorized pages - The time factor for recognition of the iics categorization code is an important variable in the usefulness of the code and, as such, the measurement of the elapsed time between placement of the categorization code and its discovery by crawlers is significant.
3. Placement of iics categorization code - For purposes of uniformity, ease of application and noninterference of the visual look of the categorized pages, it was determined that the best location for the categorization code would be at the bottom or end of the page. For this beta test the iics categorization code was typed directly onto the pages being categorized. Webmasters and others will soon be able to use an interactive interface for this purpose.
4. Initial search capabilities of the iics categorization code without the use of the full iics methodology or user interface - Pages that are categorized with the iics will become searchable via the iics categorization code on a limited but effective basis. Finding the scope of this search capability was an important aspect of the beta test.
5. Archiving potential of the iics categorization code - It is important to examine the benefits of having pages and sites that are archived with the iics categorization code in terms of the potential this offers for future retrieval. This element of the beta test is an adjunct to the first four listed since placement, discoverability and readability must first be established.
METHOD & STANDARDS USED FOR BETA TEST
Given the nature of the IICS and the global scope of its potential impact, it was decided to set up the beta test in a way that makes use of the existing infrastructure of the Internet. In particular, this IICS beta test was designed to use substantial search engine and directory resources already in place. This has not only provided us with more relevant data but has allowed us to conduct an expansive beta test at very little cost to IconFind, Inc.
The beta test began with the IICS categorization of an entire web site (www.iconfind.com ). This entailed simply typing iics categorization code at the bottom of 22 web pages which were categorized beginning on May 1, 2001 as follows:
iicsinscvi - 12 pagesPosting of all 22 pages was completed on May 3, 2001. Thereafter, at regular intervals, two of the most used search engines (Google and FAST) were checked to see if and when the iics categorization for these pages was indexed. Google and FAST were chosen for this beta test for a number of reasons as listed below:
iicsinsced - 2 pages
iicsinscte - 1 page
iicsinscfi - 1 page
iicsinsc - 6 pages
They are very popular among searchersIn addition to the above reasons, the indexes of Google and FAST are used by a number of additional search engines and directories. For instance, Google's index is used by Yahoo and Lycos utilizes the FAST index. This means that once a page has been indexed by either of these search engines, these indexes can be accessed through a much larger number of search engines and directories.
They cover a substantial number of Web pages
They update their index on a regular basis
NOTE: Regarding the time it takes for search engines to index the iics categorization code, it was decided to use the simplest approach possible for this beta test. That is, we simply entered the iics categorization code on the bottom of each page and waited for the search engines to find the pages for indexing. With some search engines and directories, webmasters could elect to have their iics categorized pages indexed more rapidly by metatagging the iics categorization code and/or by contacting those search engines and directories that accept direct submissions.
On May 25, 2001 it was discovered that both Google and FAST had indexed the iics categorization codes on IICS categorized pages. In the case of Google, all 22 iics categorization codes had been indexed. In the case of FAST, 8 categorization codes had been indexed. They break down as follows:
iicsinscvi - 12 pages ( figures 1a , 1b , 1c )
iicsinsced - 2 pages ( figure 2 )
iicsinscte - 1 page ( figure 3 )
iicsinscfi - 1 page ( figure 4 )
iicsinsc - 6 pages ( figure 5 )
iicsinscvi - 6 pages ( figure 6 )
iicsinscte - 1 page ( figure 7 )
iicsinsc - 1 page ( figure 8 )
In addition, on May 25, 2001, Lycos was also listing the iics categorization codes as indexed. As is to be expected, the results from Lycos were identical to those of FAST (figures 9 and 10 ).
Three days later on May 28 and also on May 29, Yahoo, through it affiliation with Google, began listing some of the results from the Google index ( figures 11 , 12 , 13 and 14 ).
Finally, when searches were conducted using the iics categorization code as a keyword in conjunction with another keyword as explained in the implementation plan (codes & categorization), search results were further refined and more highly relevant. For documentation of these searches see figures 15 , 16 , 17 , 18 , 19 and 20 .
An analysis of the results of this beta test as they relate to the goals previously outlined produced the following evaluations:
1. Using the iics categorization code outside the framework of programming or markup languages such as HTML or XML - By making it possible to easily place the iics categorization codes at the bottom of each page being categorized wherein they can be indexed by search engines and directories, several important objectives are accomplished. First, the categorization and coding process has been greatly simplified. Knowledge of programming and markup languages by the individual doing the categorizing are not required. This will enable people at all levels of computer skills to categorize their sites, from the beginner to the more experienced. This also means that the iics categorization code is not reliant on any software or hardware platform but can be used universally.
In addition, the iics approach does not increase the burden on search engines, directories and browsers but rather simplifies the process for them as well as the users since updates and changes in programming languages will not adversely affect the ability to read the iics categorization code. This is in contradiction to systems such as the Dublin Core and ICRA, both of which have relatively time consuming and complicated procedures based on markup languages.
2. Length of time it takes search engines to index the iics categorization code on categorized pages - All 22 of the test pages were indexed to varying degrees after an elapsed time of approximately three weeks. This time period most likely could be reduced if webmasters submitted a request for indexing to search engines and directories that accept such submissions. In any case, it is an acceptable period of time although IconFind is currently working on ways to make the iics categorization code available immediately to both search engines and directories.
Also, since the indexes of the two search engines used in this test are shared by other search engines and directories around the world, this test has shown that the iics categorization code will have widespread coverage even within the existing framework of the current search engine infrastructure.
In addition, by showing that search engines will be able to recognize and index the iics categorization codes, it is also established that these categorization codes can also be read by such search engines and incorporated into an iics database structure with minimal effort.
3. Placement of the iics categorization code - Placing the iics categorization code at the bottom of the page being categorized worked exactly as planned. It allows for uniform access to search engines, directories and browsers as well as identifies the contents of the page to those who wish to read the code visually when browsing. However, if the webmaster does not wish to have the iics categorization code seen by visitors, the code can be placed on the page using a text color that matches the background. This would render it virtually invisible without sacrificing any of its iics categorization capabilities.
NOTE: The iics categorization code will work regardless of placement on the page. However, for the reasons enumerated above, it is highly recommended that it be placed at the bottom of the page.
4. Initial search capabilities of the iics categorization code without the use of the full iics methodology or user interface - This beta test confirms that the iics categorization code can be used as a tool for refining searches using existing search engines and directories as described in the implementation plan. This capability exists in the architecture of the iics categorization code and does not need any cooperation or modification from any existing search engines and directories. Of course, as previously stated, search engines and directories that elect to use the IICS user interface and search methodology will experience a far great efficiency and relevance in their product.
5. Archiving potential of the iics categorization code - This beta test has established that the iics categorization code can be placed on pages for later retrieval and interpretation. This means that pages and sites that are archived with the iics categorization codes will have the substantial benefit of being able to retrieve the archived material through the use of the IICS user interface and search methodology.
Evidence gathered and analyzed during the course of this beta test firmly establishes that the iics categorization code can be easily placed on web pages as a simple text string outside the framework of any programming or markup language. It can then be retrieved through the automatic crawler process of search engines.
In addition, the categorization code can initially be used for searches within the existing search engine and directory framework. Search engines and directories, upon licensing, may also elect to use the full iics search methodology and user interface for much greater effectiveness in content organization and retrieval.
Finally, use of the iics categorization code on pages and sites will lead to more efficient retrieval of this material once it has been archived.
June 22, 2001