Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
1999-10-28
2001-02-27
Amsbury, Wayne (Department: 2771)
Data processing: database and file management or data structures
Database design
Data structure types
C707S793000
Reexamination Certificate
active
06195652
ABSTRACT:
BACKGROUND OF THE INVENTION
This invention relates to the field of databases that index large numbers of different types of items.
BACKGROUND
The Internet is by now the world's largest computer network, interconnecting millions of computers. One side effect of this large size is that the vast amount of information available on the Internet is often extremely difficult to access. Similar problems tend to occur on any large network, and in this discussion the Internet is used merely as an example of such a network.
Several attempts have been made to index the Internet, but each of these attempts has fallen short. Probably the most primitive indexing scheme comprises a simple alphabetical indexing of items, such as URLs in the case of the Internet. Unfortunately, such an index is little more than an electronic “white pages,” and is generally only useful if one knows the exact name or address of the user to be contacted. A straight alphabetical index also provides little or no ability to access information by subject matter, unless the name happens to include an indication of the subject matter, as in “Bob's Pharmacy.” These shortcomings are quite annoying when the number of addresses being indexed is relatively small, and present profound difficulties as the number of URLs grows.
Another method of indexing information involves keyword searching. In keyword indices information is contained in freeform text fields, and then the system, the system manager, or the user indexes key words found in the text. A considerable amount of scientific information is presently indexed in keyword indices such as Medline™, and literally millions of web pages of the Internet are indexed by such services as GOTO™ and Lycos™. Although advantageous relative to simple alphabetical address indices, keyword indices are still impractical where the amount of information is relatively high.
One major problem is that keyword searching is inherently under-inclusive. For example, if one is looking for a red colored automobile, a keyword index would only be useful to the extent that all red cars arc indexed using the color red. Cars indexed as rose or magenta, crimson, ruby, vermilion, or scarlet would not be located by a search for “red.” Keyword searching is also inherently over-inclusive. A search for a red Mercedes™ may well identify a story about a woman named Mercedes who is wearing a red dress. There are presently attempts being made to provide proximity and content-based keyword searching, including the use of XML standards, but such attempts have so far been seen only marginal benefit. Even if keyword searching could be utilized to carefully exclude irrelevant information, it would still tend to be under-inclusive. Moreover, keyword searching of any degree of sophistication inherently suffers seriously diminishing utility as the number of records increases. For example, it does little good to conduct a search for a desired automobile if the database locates 5000 records, and there is no practical way of further limiting the search without rejecting desired records.
A great deal of information on the Internet is now stored in specialty databases. In specialty databases the subject matter is limited to a particular type of product or service, and parameters are provided which specifically relate to that subject matter. Thus, a specialty real estate database may store data according to the parameters of property type (commercial, residential, undeveloped land, etc.), location, square footage, number of bedrooms, number of baths, style (Mediterranean, Cape Code, traditional, etc.), and price. Additional information may also be stored in a comment or image field, but such information is typically not searchable or sortable, and is difficult to standardize. In addition, specialty databases are unable to properly handle multiple types of items, i.e., items which are generally described using inconsistent or otherwise different parameters. Thus, a real estate index is very poorly adapted for storing information relating to clothes or automobiles because the data parameters required for these items arc almost completely inconsistent with the parameters useful for real estate.
One technique that has proven useful for organizing information relating to multiple types of items in a generic index is a hierarchical “yellow pages.” Thus, for example, the Big Book™ collects together Internet addresses for various categories such as attorneys, bookkeepers, florists and so on. The Big Book™ also indexes Internet addresses according to geographical location. Thus, one can select family law attorneys in Miami, rather than all attorneys, or all family law attorneys throughout the country. Despite these advantages, the “yellow pages” type of index is still not particularly useful where the subject matter one is searching for is not well categorized, or where the categories are vague or counterintuitive. For example, if one is searching for zipper manufacturers, a “yellow page” type index may not have any categories specifically for zippers, or zippers may be listed under some obscure category such as “interlocking clothing connectors.” Still further, such indices are notoriously cumbersome when searching for related products. For example, if one is searching for all types of clothing connectors, a “yellow pages” type of index may require separate searches for zippers, buttons, snaps, etc.
Similar problems exist with respect to limiting the searches by geographical location. For example, an index may well have categories for continents, countries, major metropolitan markets, or even specific cities, but a user must still discover on his own how the database treats a particular address such as “upstate” New York. In short, unless a user knows how the index is organized, it may be nearly impossible to find desired information in a convenient manner.
A better solution for organizing information in a generic index involves hierarchical sorting of products and services as is done by Netscape™ and others. In a hierarchical index one can begin with a high level category, and then logically work down to a lower level category by selecting choices from various menus. Thus, in selecting patent attorneys one might sequentially choose the following categories: Services (level 1), Business (level 2), Legal (level 3), Attorneys (level 4), and finally Patent (level 5). The same strategy can also be applied to geographical locations. One may, for example, select Los Angeles by choosing North America (level 1), United States (level 2), California (level 3), Southern California (level 4) and Los Angeles (level 5).
However, even generic indices that select records hierarchically are unsatisfactory for accessing huge amounts of information. One problem is that presently known hierarchical indices do not allow users to adequately filter down to smaller subsets of records, or to sort the selected records based upon parameters having particular meaning for the field being searched. A search in the field of automobiles can be used to illustrate these shortcomings. If one searches a presently known Internet index for used Ford™ automobiles, one might locate thousands of cars for sale. Clearly that number is too large for a user to realistically review every listing. Even if the number of selected records could be reduced by limiting the subject matter to a particular model, or by limiting the geographical location to a particular city, there might still be several hundred cars for sale, and it would be a terrible waste of time to consider each of those records if the user is only looking for cars with purchase prices of less than $3000. Specialty indices may allow a user to limit the search by price, color, body style, and other parameters all at the same time, but that capability is not available on generic indices which additionally cover other types of items, such as real estate and computers.
Even if generic indices could be developed that use different sets of parameters for different products, the creation costs of such an index wou
Amsbury Wayne
Fish Robert D.
Fish & Associates, LLP
LandOfFree
Self-evolving database and method of using same does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Self-evolving database and method of using same, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Self-evolving database and method of using same will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2563975