GB2366408A

GB2366408A - Compatibility selection system

Info

Publication number: GB2366408A
Application number: GB0021529A
Authority: GB
Inventors: Jose Antonio Guerrero
Original assignee: Individual
Current assignee: Individual
Priority date: 2000-09-01
Filing date: 2000-09-01
Publication date: 2002-03-06
Anticipated expiration: 2020-09-01
Also published as: GB0021529D0; GB2366408B

Abstract

An algorithm and system that permits the selection of the most compatible goods or services from suitable computer databases; extreme results, like zero or thousands of 'hits', do not occur. The system selects Objects, like personal computers or job vacancies, stored in object databases; the objects selected are the most compatible with a Subject (or Model) defined by the consumer; the Models are stored in subject databases. For Objects and Subjects appropriate questionnaires collect characteristics (descriptions) that are captured as intensities (weighting factors) of human feelings expressed as strings of digits. A computer program operates on the strings of digits to perform, in sequence, sumproducts (sums of products) between Subject and Object intensities. Objects with the highest sumproduct Score are selected; the degree of compatibility, in percent, is measured. In some applications, like selecting a house, the selection establishes the highest one-way (uni-directional) compatibility between the Subject and Objects; in other applications, like job search, the selection establishes the highest two-way (reciprocal) compatibility between the Subject and Objects. The flexible algorithm permits efficient implementation on many applications on the Internet as well as stand-alone digital computers.

Description

<Desc/Clms Page number 1> COMPATIBILITY SELECTION SYSTEM TECHNICAL FIELD This invention relates to the compatibility selection of objects from a database by weighted sumproducts where the object selected can be a person, thing or service. BACKGROUND OF THE INVENTION It is common in the information technology (IT) industry to gather actual data into a pool or database and then attempt to select objects in a controlled manner by defining comparison models as the basis for selection. The selection is accomplished by comparing the actual objects with the described models and selecting the objects matching the models.

I list below some applications of systems for the selection of objects: 1) Selection of personal computers (PC's) to buy 2) Selection of dating or marriage partners 3) Selection of books/publications to buy 4) Selection of jobs wanted (recruitment) 5) Selection of houses to buy 6) Selection of cars, boats, ships or planes to buy 7) Selection of companies for mergers, acquisitions 8) Selection of patents, etc.

It is in order to state that if a consumer wishes to select a personal computer to buy he will first describe the ideal characteristics (the Model) of the PC he is interested in buying. The description of this model PC will be compared with descriptions of many actual PC's for sale present in a suitable database. The selection system then attempts to identify the actual computer(s) matching the model PC described by the consumer.

Current shortcomings Current selection systems (i.e. Internet search engines) have some of the following shortcomings: 1) For different selection characteristics (i.e. key words), no efficient provisions to give relative weights to the different charac -ristics in accordance with their importance.

2) For characteristics containing several options/choices, no efficient provision for the consumer to indicate degrees of preference (human intensity) for the different options.

3) Logical (Boolean) and/or mathematical operators usually control the selection of objects, which at times produces zero results or thousands of results (hits).

4) No provision to measure degrees of compatibility and therefore no assessment of the quality of the database as related to particular selection attempts.

5) No reciprocal matching (i.e. great if the job on offer matches the applicant's requirements, but do the applicant's qualifications matches the employers' requirements?) 6) No generalised computer source code for different selection applications (i.e. drastically different source code architectures are used for different selection applications). Analogy I will use an analogy to illustrate the above problems.

Suppose you have an acquaintance, Mr. Smith, with whom you are out of touch for several years. You now wish to find Mr. Smith's telephone number in your local phone book. If you remember his precise name of G.O.D. Smith then you will find his phone number immediately. If Mr. G.O.D. Smith moved away 3 years ago then your search produces zero results. If you remember only his surname, Smith, then you could be confronted with hundreds of listed persons (hits) with the surname Smith. If you remember

.iis surname Smith and his street address then you could use these two pieces of information (equivalent to a Boolean search operation) to painstakingly search for your acquaintance's phone number. However, if you don't remember Mr. Smith's surname and you remember that he is 6-ft tall, with red hair, about 40 years old, light complexion, non-smoker, left handed, blue eyes, a mechanic, married and with 2 children then currently you cannot use this additional information to find your acquaintance's phone number. You would need to have the additional information in a suitable telephone database and you would also need an appropriate computer program to do a rigorous compatibility search.

The information technology (IT) industry has not yet produced a very practical manner to capture and retrieve the immense number of pieces of information needed and the computer program to perform an accurate and efficient compatibility search of the results. The compatibility approach is specially needed if the object database is updated regularly and by now Mr. G.O.D. Smith is really 45, has been busy making a third child and moved house one mile away. If there is no compatibility approach then the search result would probably be zero. My compatibility system would detect Mr. G.O.D. Smith's telephone number without having to know his exact name or address.

I do not maintain that my compatibility selection system could or should substitute current telephone search systems - I am only illustrating a different approach to be used in more relevant and demanding commercial applications. Illustrating the problems I will illustrate the problems mentioned above with some practical examples as described below.

Medical Diagnosis It is estimated that up to 40,000 people a year die in Britain from errors in medical diagnosis (Lois Rogers, "Blunders by doctors kill 40,000 a year", The Sunday Times, 19 December 1999). In an effort to modernise the National Health Service (NHS) the Government of the United Kingdom has introduced the website "www.nhsdirect.nhs.uk" for online medical diagnosis. In essence, this NHS website will identify the symptoms if the disease has already been diagnosed - a simple listing. But this information is readily available from any medical encyclopaedia. Applications like the NHS diagnosis do not really exploit the computational power of the digital computer. It would be more beneficial if the NHS diagnosis system were to, given the symptoms, identify possible diseases. A medical doctor's confirmation of the sickness would then be appropriate.

Note that in the NHS website a picture of the human body represents the main database and parts of the body represent smaller sub-databases; this approach to "divide and conquer" is a recommended approach in handling some large databases (i.e. initial classification of books into subjects). Listings similar to the NHS diagnosis can be obtained from "www.webmd.com" under 'Health Topics A-Z'. Searching for Patents Definitions of the mathematical concepts of autocorrelation and crosscorrelation are found in the book "Encyclopaedic Dictionary of Exploration Geophysics" by Robert E. Sheriff, ISBN: 1560800186. Autocorrelation and crosscorrelation techniques have been used in many inventions in the oil- exploration, geophysical and seismic industries. US Patent 5,838,564 is a relevant example in the use of autocorrelation and crosscorrelation techniques. My Boolean search for terms autocorrelation-AND-crosscorrelation in "vww.uspto.gov" (a11

years, all fields) produced 151 hits, not including patent 5,838,564. This is too many hits to be considered an efficient search, especially when a good example is missing from the results. A search system that would permit me to describe more than two key terms would improve the efficiency in my quest to find patent 5,838,564 or any other relevant patent. In this particular case, my suggestion of some additional terms would be seismic, autocorrelation, crosscorrelation, 3-D, geophones, computer, coherency, traces, etc. Searching for Books Amazon Books sells books via the Internet "www.amazon.com" and claims to have the "Earth's Biggest Selection". To buy a book from Amazon you need to know the name of the book or the name-of the author or the ISBN number. For example: "Applied Mathematics for Business, Economics, and the Social Sciences" by Frank S. Budnick, Published by McGraw-Hill Inc., ISBN: 0071125809. If you do not have this information then you are confronted with searching by subject, in this case, 'mathematics'. My search in "www.amazon.com" for the subject 'mathematics' produced 32,000 hits, not a very helpful piece of information indeed. I wanted to buy a book with seven maths topics that interested me, as follows: 1) Linear Equations 2) Exponential and Logarithmic Functions 3) Matrix Algebra 4) Linear Programming 5) Probability Theory 6) Differential Calculus 7) Integral Calculus.

I sent an email to Amazon Books (Appendix 1) requesting help in searching for a book with the above seven topics. Amazon Books replied (Appendix 2) as follows: "Unfortunately, please

. < now t_r:at we do not currently have a way to search by topics or table of contents. However, I think this is a grew- suggestion, and I've passed it on to the appropriate department in our company for consideration. We truly value this kind of feedback, as it helps us to continue to improve our store and provide better service to our customers". As it happens, the seven maths topics listed above are chapters in Budnick's book, and Amazon's search engine currently is not able to select Budnick's book, or any other compatible book, by interrogating its table of contents or topics. Journalists, researchers, specialist bookstores, universities, public libraries, patent offices and other institutions could benefit with an alternative selection system by book topic. Searching for Social Partners (dates) Some people sneer at computer dating but to engineer a sophisticated computer dating system is very difficult because it involves acknowledging too many subjective characteristics such as likes, dislikes, tastes, feelings, personalities, preferences, habits, behaviours, beliefs, etc. If a good computer-dating system can be designed then such system, if easily modified, can be used in many other applications. In recent years there has been an explosion in computer dating firms operating in the Internet. The Aarens Dating Directory lists two pages of Match Maker Sites with 148 dating organisations. This directory can be accessed as follows: "www.aarens.com"_1o- Dating Services Directory__p# Matchmakers From Page 2 of the Matchmakers Directory I have selected two dating firms that permit us to examine their questionnaires: 1) BestMatch: "www.bestmatch.com" (Appendix 3) 2) MatchMaker: "www.matchmakerintl.com" (Appendix 4) An examination of these questionnaires reveals that, in general terms, there is very limited compatibility matching since no entries are available for degrees of preference

(intensities). Also, the questionnaires are not rigorous enough to permit reciprocal matching, a basic requirement for good quality computer dating. My compatibility selection system is more robust than most of the systems listed in the Aarens Dating Directory. Tender Buying In The Sunday Times newspaper of London of the 25 of July 1999, the columnist Mr. Matthew Lynn wrote the profile "Fighter pilot shoots from the hip" about the founder and chairman of the computer company Oracle, Mr. Larry Ellison. In this profile Mr. Lynn wrote about one of Mr. Ellison's predictions: "... Right now, his prediction is of the world turning into a giant electronic auction house, with every transaction leading to a series of bids made over the Internet. His example is of a company buying 1,000 new desks. Instead of asking a few local stores for a price, it would post what it needed on the Internet and companies around the world would automatically tender for the business. Essentially, every market would become a miniature version of the stock or currency markets..." My invention can convert Mr. Ellison's business-to-business (B2B) prediction about tender buying into a practical reality. Inadequate Software On the Atlantic edition of TIME Magazine, issue dated 3 August 1998, an article named "Click Till You Drop" about the Internet economy was published. The article states: "...Yahoo charges about 4 cents for every ad it serves up on many of its 115 million pages every day. And those prices will rise as Yahoo develops technology that lets it more closely match advertisers with searchers..." This statement reflected the situation that, two years ago, software developers had not yet developed rigorous compatibility search engines. I believe that at present

-his is still the case - that current search engines need to improve their selection techniques in order to provide a better service to the consumer.

On `he European edition of FORTUNE Magazine, issue dated 4 of SeptetmDer 2000, an article named "Toil and Trouble: Online Shopping Is Still a Muddle" states that: "...just 58$ of the (50 Internet) sites had a search function that really did its job..."; a statement also appears that reads: "... Online computer sellers have their own problems. Features like the ability to customise a computer system online are of little use when the site doesn't provide any guidance on what computer components and software to select..." I also believe that my compatibility selection system can provide such needed guidance and completeness not just for the sale of PC's but in many other applications as well. Internet Profits It is estimated that ninety percent of the Internet companies are losing money (Edward H. Amory, "Day the Internet grew up", London Daily Mail, 12 January 2000). One of the reasons for this lack of profits is because most of the search engine companies provide results free of charge - do not charge directly to the consumer for search services. Most Internet companies derive their income from indirect advertisement revenue but there is a practical limit to the availability of advertisement accounts. Most Internet companies do not risk to charge directly to the consumer because the quality of their service is not optimum. My invention offers optimum selections with the option for the consumer to inspect results before choosing to pay for the results.

BRIEF SUMMARY OF THE INVENTION.

It is the objective of the present invention to provide a new and improved data processing system in which a digital computer is used for the compatibility selection of objects, such as goods or services, by comparing the Objects' characteristics (description) with a Subject's (the Model's) characteristics as described by a consumer, my system comprising the steps of: means of completing intensity questionnaires for receiving numerical values assigned to the characteristics as expressions of human preference; means for using a digital computer responsive to such numerical values for generating intensity databases by incorporating the information entered on the questionnaires; means for using a digital computer responsive to the intensity databases for the one-way or two-way (reciprocal) compatibility selection of the Objects, by weighted sumproducts with the Model; means to measure in percent the degree of compatibility between the selected Objects and the Model; and display means for exhibiting the results of the selections.

Let us suppose that you wish to search for an object like a house, a computer, a date, a job, etc. My system, and most commercial search systems, will deliver a selection if there is a good match in the databases. The question is: what happens if there is no match? Most commercial operators deliver an unhelpful message like "no matches available". On the other hand, my system always delivers a solution - it delivers a few selections that are the most compatible with the described model. The quality of my selections is directly related to the quality of the databases. There is also the other extreme in which commercial operators allow only one or two key words as search criteria, the result being hundreds or thousands of selections (hits), result that brings little comfort to the

,:onsumer. The intention of this compatibility selection system is to perform, in a simple but efficient manner, rigorous selections of data stored in computer databases. The selection is done on a compatibility basis between a Model and actual objects. It is done by capturing, in practical and organised questionnaires, detailed information describing Models and objects such as goods, people, services, etc. to be stored in suitable databases. The information is stored as 'strings' of digits suitable for their mass processing in devices like digital computers. The arrangement of the captured data permits the effective use of the mathematical technique of weighted sumproducts that determines the degree of compatibility between two different sets of data. This leads to the selection of objects with the highest compatibility scores. The Score corresponds to a relative measure of compatibility. An additional measure of the degree of compatibility in percent between the Model and the selected objects is performed and it is named the Compatibility Factor. Hand computations are also possible but this becomes a laborious task.

Advantages Compared to the current selection systems (i.e. Internet search engines) my invention has the following advantages: 1) For different selection characteristics (i.e. several key words), efficient provisions to give relative weights to the different characteristics in accordance with their importance.

2) For the several options contained in each characteristic, efficient provision for the consumer to indicate degrees of preference (human intensity) for the different options.

3) Provision for the measure of relative compatibility by sumproduct scores. The selection of objects is controlled by sumproduct scores, which do not produce zero results or thousands of results (hits).

4) Provision to measure the degree of compatibility in percent (Compatibility Factor) and therefore an additional assessment of the quality of the database as related to particular selection attempts.

5) Reciprocal matching (i.e. the job applicant's requirements are interrogated against the jobs on offer and simultaneously the applicant's qualifications are interrogated against the potential employers' requirements).

6) A generalised computer source code for different selection applications (i.e. drastically different source code architectures are not needed for different selection applications). Generalised source code is currently in demand by Application Service Providers (ASP). Simulations In my Detailed Description of the Invention I demonstrate the workings of my invention by simulating two commercial applications. The first simulation is for the sale of personal computers (PC's), simulation that uses the one-way compatibility selection approach. The second simulation is for computer dating, simulation that uses the reciprocal (two-way) compatibility selection approach. The results produced by these two working models are corroborated with detailed spreadsheet calculations. Solutions to problems I will now further discuss the advantages of my invention by proposing solutions to some of the problems identified in my Background of the Invention. Medical Diagnosis Using my invention it is feasible to identify possible diseases after completing compatibility questionnaires

describing the symptoms. To my knowledge, no current search system is properly diagnosing diseases. The workir-,g models described in my Detailed Description delineate th: practical framework for implementing medical diagnosis by compatibility selections. Searching for Patents Performing a c:-e-way compatibility selection as follows can solve the problem of searching for relevant patents in 'www.uspto.gov': a) The patents are classified into fields like the fields of physics, chemistry, mechanics, electronics, seismology, biology, computers, etc. Each field would require its own object database and its own subject database.

b) Suitable intensity questionnaires are constructed for each field. The questionnaires must accurately describe the characteristics and options relevant to each field.

c) A questionnaire is completed for each patent in the field by entering intensities for each relevant option. The intensities are entered as a string of digits in a computer object database. Each patent is given an object identification code number.

d) The consumer/user, in his own questionnaire for the field, enters intensities specifying his selection criteria (the Model) in a similar manner as the completed patent questionnaires. His selected intensities are entered as a string of digits in a subject database. The User is given a subject identification code number.

e) A 'sumproduct' for two strings is the product of an option's intensity value in the first string (object) times its corresponding option's intensity value in the second string (subject) followed by the sum of a11 similar products obtained between the two strings. The resulting total sum I call the

'Score'. For a particular field, the system's computer program performs sequential sumproducts between the Model's string and each of the strings for the patents in the object database. Weights are introduced if so desired. Code numbers for patents with the highest weighted sumproduct scores are saved as desired selection results. These scores thus represent the relative measure of compatibility.

f) After the selection of compatible patents, the program performs, on the Model's string, an auto-sumproduct for each characteristic in the Model. The program has already obtained scores between the Model and the selected patents. Both of these results are used to compute Compatibility Factors expressed as percentages. The Compatibility Factor (CF) is a measure of the degree of compatibility in percent between the subject (the Model) string and each of the selected strings from the object database (selected patents from the field).

g) The code numbers are needed for the internal workings of the program. The code numbers provide the cross reference of the User's Model with the patents selected as most compatible with the User's Model. The use of code numbers also provides the option, if so desired, to charge a fee to the User before the release of the patent's identity. Interested persons looking for relevant patent information would be willing to pay extra for this effective additional service. A User or Patent Examiner, considering a patent application, could expediently learn if a particular idea has already been patented or not.

h) A result sheet, listing the options with the highest intensity values (primaries) for the selected patents' code numbers, is presented to the User for him to decide whether or not to pay for the release of the identity of the patents selected - this is an innovative and significant incentive to electronic commerce.

i) I refer to this type of selection as a 'one-way'

,ompatibility selection because the sumproducts are performed between only two different intensity strings. One string is the intensities specified by the User (the Model) and the second string is the intensities entered for each patent in the field. Searching for Books The problem of not being able to buy books by specifying the topics of interest can be solved using the same procedure as the procedure to select patents as mentioned above. Books are normally searched by 'subject' but this term would introduce confusion with my term 'subject database'. Hence I will use the term 'field' to classify books, as I have already used the term 'field' to classify patents. I will describe below the procedure to select books by topics performing a 'one-way' compatibility selection: a) The books are classified into major fields, like the fields of physics, economics, medicine, chemistry, mechanics, electronics, mathematics, seismology, biology, computers, accounting, etc. Each field would require its own object database and its own subject database.

b) Suitable intensity questionnaires are constructed for each field. The questionnaires must accurately describe the characteristics and options relevant to each field, reflecting in a generalised manner the Table of Contents of the books in the field.

c) A questionnaire is completed for each book in the field by entering intensities for each relevant option. The intensities are entered as a string of digits in a computer object database. Each book is given an object identification code number.

d) The User completes his own questionnaire for the field specifying his selection criteria (the Model book) in a similar manner as the completed book questionnaires. His selected intensities (Model) are entered as a string of digits in a

subject database. The User is given a subject identification code number.

e) For a particular field, the system's computer program performs sequential weighted sumproducts between the Model's string and each of the strings for the books in the object database. Code numbers for books with the highest sumproduct scores are saved as desired selection results.

f) After the selection of compatible books the program performs, on the Model's string, an auto-sumproduct for each characteristic. The program has already obtained scores between the Model and the books selected. Both of these results are used to compute Compatibility Factors expressed as percentages.

g) The use of code numbers provide the option, if so desired, to charge a fee to the User before the release of the actual book's identity. Interested persons looking for relevant book information would be willing to pay extra for this effective additional service.

h) A result sheet, listing the options with the highest intensity values (primaries) for the selected books' code numbers, is presented to the User for him to decide whether or not to pay for the release of the identity of the books selected - this is an innovative and significant incentive to electronic commerce. Searching for Social Partners (Computer Dating)) As I mentioned before, I demonstrate the workings of my reciprocal (two-way) compatibility selection technique with a simulated commercial application (a working model) of computer dating . This simulation clearly demonstrates the advantages of my approach, especially the reciprocal matching innovation, as compared to the shortcomings of the agencies currently offering computer-dating services as mentioned before. Reciprocal compatibility selections can also be effective in executive

recruitment, adoption of children, company mergers, mortgages, registration to university courses, joint-ventures and other applications that require two-way interrogation. Tender Buying Let us consider the case of B2B tender buying. Mr. Larry Ellison predicts that sometime in the future a company will be able to buy 1000 desks by posting in the Internet tenders for the purchase of the desks . If you wish to purchase a book from Amazon Books, it is Amazon Books (the seller/provider of the goods) that sets up and manages the search system (i.e. questionnaires, databases, website, etc). In tender buying it would be the buyer who would set up and manage the selection compatibility system. The buyer would produce the relevant intensity questionnaires to accommodate the technical specifications for the desks and the terms and conditions for the purchase. The intensity questionnaires could have scores, hundreds or even thousands of options if necessary. The buyer would complete his own intensity questionnaire specifying his ideal purchase (his Model). Sellers around the world would submit their tenders by completing their own intensity questionnaires, one for each desk model manufactured, via the Internet, Fax, post, etc. Using my compatibility selection system the buyer could then select, in a few seconds, the most compatible tenders to his purchase Model, changing Mr. Ellison's prediction into a reality. In tender buying it would be simple to request tenders for the purchase of, say, 5000 automobiles by a car rental company, or an executive aeroplane, 1000 personal computers, a castle in Europe, an ocean going yacht, etc. and select very expediently the best sellers. My compatibility selection system is therefore flexible because it can be used to buy or to sell goods and services.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 illustrates a Characteristics example. FIG. 2 illustrates an Options example.

FIG. 3 illustrates an Intensities example. FIG. 4 illustrates a Sumproduct example.

FIG. 5 illustrates an Auto-Sumproduct example.

FIG. 6 illustrates a Compatibility Factor (CF) computation. Figs. 7A and 7B illustrate a sample selection of personal computers (PC's).

FIGS. 8A through 8C illustrate a blank General Intensity Questionnaire for personal computers.

FIGS. 9A through 9C illustrate a completed PC Buyer's Intensity Questionnaire.

FIGS. 10A through 10C illustrate a completed PC Seller's Intensity Questionnaire.

FIGS. 11A and 11B illustrate Run A - a selection of PC's using primary intensities with no scalars.

FIGS. 12A through 12C illustrate, for Run A, the Sumproduct between buyer B0009500 and seller S0000378 using primary intensities with no scalars.

FIGS. 13A and 13B illustrate Run B - a selection of PC's using primary intensities with scalars.

FIG. 14 illustrates the scalars used in selecting PC's. FIGS. 15A through 15C illustrates, for Run B, the Sumproduct between buyer B0009500 and seller S0000450 using primary intensities with scalars.

FIGS. 16A through 16C illustrate, for Run B, the Auto- Sumproduct for buyer B0009500 using primary intensities with scalars.

FIGS. 17A through 17D illustrate, for Run C, the selection of PC's using primary and secondary intensities with scalars. FIGS. 18A through 18C illustrate, for Run C, the Sumproduct

oetween buyer B0001500 and seller S0000491 using primary and secondary intensities with scalars.

FIGS. 19A through 19C illustrate, for Run C, the Auto- Sumproduct for buyer B0001500 using primary and secondary intensities with scalars.

FIG. 20 illustrates, for buyer B0001500, the summary of results of Runs A, B and C.

FIGS. 21A through 21F illustrate the perfect match for buyer B0009491 (primaries only) with compatibility factor of 1000. FIGS. 22A and 22B illustrate the perfect match for buyer B0000491 (primaries and secondaries) with compatibility factor of 34.77%.

FIG. 23 illustrates the reverse selection for seller S0000491. FIG. 24 illustrates the repetition of Run C changing the Monitor scalar to 550 and the Approximate Price scalar to zero.

FIGS. 25A and 25B illustrate a sample selection of computer dating (for a male client).

FIGS. 26A through 26E illustrate a blank Computer Dating Intensity Questionnaire.

FIGS. 27A through 27E illustrate a completed Computer Dating Intensity Questionnaire for male M0000015.

FIGS. 28A through 28E illustrate a completed Computer Dating Intensity Questionnaire for female F0000010.

FIGS. 29A and 29B illustrate Run D - a selection of dates using primary intensities with no scalars.

FIGS. 30A through 30F illustrate, for Run D, the Sumproduct between male M0009015 and female F0009010 using primary intensities with no scalars.

FIGS. 31A and 31B illustrate Run E - a selection of dates using primary intensities with scalars.

FIG. 32 illustrates the scalars used in selecting social partners (dates).

FIGS. 33A through 33F illustrate, for Run E, the Sumproduct

between M0009015 and female F0009010 using primary intensities with scalars.

FIGS. 34A through 34F illustrate, for Run E, the Auto- Sumproduct for male M0009015 using primary intensities with scalars.

FIGS. 35A through 35D illustrate, for Run F, the selection of dates using primary and secondary intensities with scalars. FIGS. 36A through 36F illustrate, for Run F, the Sumproduct between male M0000015 and female F0000011 using primary and secondary intensities with scalars.

FIGS. 37A through 37F illustrate, for Run F, the Auto- Sumproduct for male M0000015 using primary and secondary intensities with scalars.

FIG. 38 illustrates, for male M0000015, the summary of results of Runs D, E and F.

FIGS. 39A and 39B illustrate the reverse selection for female F0000011.

FIG. 40 illustrates a graphical representation of the compatibility technique. FIG. 41 illustrates the Sumproduct scores for the graphical representation of the compatibility technique of FIG. 40. FIG. 42 illustrates the Sumproduct of an unusual score result. FIG. 43 illustrates the Auto-Sumproduct of an unusual score result. FIG. 44 illustrates the Compatibility Factor computation of an unusual score result. FIG. 45 illustrates the Sumproduct of an unusual Compatibility Factor result. FIG. 46 illustrates the Auto-Sumproduct of an unusual Compatibility Factor result. FIG. 47 illustrates the computation of an unusual Compatibility Factor.

FIGS. 48A through 48D illustrate the seller's input text file "CompSell.txt" listing a Master String and record S0000378.

FIG. 49 illustrates the TITAN1 flowchart used to build databases for buyers and sellers of PC's.

FIGS. 50A and 50B illustrate the text file "PRINT1.txt" listing part of the sellers' database corresponding to S0000378. FIG. 51 illustrates the TITAN1 flowchart used to build male and female databases in computer dating.

FIG. 52 illustrates the TITAN2 flowchart used for the selection of personal computers.

FIG. 53 illustrates the file "Card2.txt" containing the Run Time Parameters (RTP) used during the execution of TITAN2 for the selection of personal computers, with and without scalars. FIG. 54 illustrates the TITAN2 flowchart used for the selection of social partners (computer dating).

FIG. 55 illustrates the file "Card2.txt" containing the Run Time Parameters (RTP) used during the execution of TITAN2 for the selection of social partners (computer dating), with and without scalars.

DETAILED DESCRIPTION OF THE INVENTION.

Arrangements according to the invention will now be described by way of examples. The first example is related to the selection of personal computers (PC's). The second example is related to the selection of social partners usually referred to as 'computer dating'. Before describing the examples in detail it is in order to describe the terminology as used in the context of my invention and common to both examples. Terminology Object - In database computer applications, an object can be a picture, chart, text or any other form of information that you create and edit. In this invention, an object can be a person, animal, article, service, thing or any other form of information stored in a suitable object database. The purpose is to select one or more objects from this object database in accordance with a selection criteria (a Model) specified by a Subject (see below). We can normally equate an object to a seller, provider, advertiser, suitor, etc.

Subject - In this invention, a subject can be a person, animal, article, service, thing, or any other form of information stored in a suitable subject database. This subject information is used as a selection criterion (a Model) to perform selections on a separate database containing the objects. We can normally equate a subject to a buyer, client, consumer, host, user, searcher, etc.

Compatibility Selection - The purpose in my invention is to select one or more objects (resident on the object database) by measuring the compatibility between a subject (a Model, resident on a subject database) and the objects. Objects with the highest compatibility with the model will be selected.

One-Way Compatibility Selections - In the case of selecting

nouses, computers, airline tickets, shares, books, patents, etc. the compatibility is measured by comparing the Model, in the subject database, with the objects in the object database. The selection criterion specified by a person looking for a house to buy (the buyer) is compared with the descriptions of the available houses for sale on the object database. This examination for compatibility is only 'one-way' because there is no need to examine if the houses are compatible with (approve of) the prospective new owner..

Two-Way (Reciprocal) Compatibility Selections - In the case of select g jobs, social dates, joint-ventures/mergers, child adoptions, mortgages, personal loans, etc. the examination for compatibility is measured twice. For example, in the case of computer dating, a 50-year-old male requests dating 20-year-old females. The selection procedure would identify one or more 20- year-old actual females, compatible with his ideal 20-year-old female Model, as described by the 50-year-old male in his own questionnaire. But would the selected 20-year-old females be interested in meeting the 50-year-old male? In a 'two-way' (reciprocal) compatibility selection, a11 the females' male Models are also measured for compatibility with the actual 50- year-old male. The female dates for the 50-year-old male are selected by measuring the compatibility between the male's ideal female Model with a11 the actual females and by simultaneously measuring the compatibility between all the females' ideal male Models with the actual 50-year-old male. The combined compatibility measurements produce a unique reciprocal compatibility selection score. In two-way compatibility selections the roles of the Subject and Object become somewhat interchangeable.

Characteristics - The characteristics are the attributes that define an object or subject. Some of the characteristics that define a person are: age, height, race, occupation, taste in

food, etc. I illustrate in FIG. 1 the characteristic 'Taste in Food' as it would be presented in a questionnaire constructed for a relevant application. Some of the characteristics that define a personal computer (PC) are: hard-drive size, processor speed, screen size, RAM size, etc. Some of the characteristics that define a house are: location, price, number of bedrooms, age, garden size, etc.

Options - Each of the above characteristics is subdivided into suitable options that further define the characteristic. A characteristic like "Taste in Food" can be subdivided into the options: German-food, Italian-food, Greek-food, etc. as illustrated in FIG. 2. A characteristic like "Screen Size" can be subdivided into the options: 13-inch, 15-inch, 17-inch, 19- inch, etc. A characteristic like "Number of Bedrooms" can be subdivided into the options: 1-bedroom, 3-bedrooms, 5-bedrooms, etc.

Intensity - I describe below my system of expressing 'intensities of human feeling' by assigning numerical values to the options described above. I use the range of numbers from +127 to -127 to indicate the order of preference, or intensity, for the above options. The use of 3-digit options would require the use of 3-digit intensities; the use of 2-digit options would. require the use of 2-digit intensities; and the use of one-digit options would require the use of one-digit intensities. For example, using one-digit intensities with a subset range of values from +9 to 0, a '9' would indicate the highest intensity and a '0' the lowest intensity or preference. In the case of the characteristic "Taste in Food", if I indicate the intensities of the above options as illustrated in FIG. 3, the intensity of 9 indicates that my highest preference is for Italian food. The intensity of 5 indicates a medium preference for Chinese and English food. The intensity of 0 indicates my dislike for Mexican food. All the other intensity values express

intermediate preferences for the other options in the characteristic "Taste in Food". Similar 'strings' of intensities are also used for the detailed description of goods, services, people, etc.

Intensity Permutations - The maximum number of possible permutations for the digits entered as intensities depend on the total number of options and the range of intensities. For example, if we construct a questionnaire to describe personal computers as follows: Number of Characteristics = 12 Number of options per characteristic = 8 Total number of options = 12x8 = 96 Range of intensities, from 9 to 0 = 10 Permutations = 10 raised to the power of

I can therefore state that the general formula to compute the maximum number of intensity permutations is:

Total Options Number of Intensity Permutations = (No. Intensities) Note that a simple questionnaire, with 12 characteristics and 8 options per characteristic and 10 intensities per option, results in thousands of millions of permutations.

Primary Vs Secondary Options - As an example, if I was to define my ideal partner's characteristic "Taste in Food" using a table like FIG. 3, normally there should be only one option associated with my true "ideal" partner's taste in food - the option with the highest intensity such as Italian food. For any particular characteristic, I refer to the option with the highest intensity as the Primary Option. In FIG. 3 the primary option would be "Italian food" as 9 is the highest intensity for

the characteristic. A11 other remaining options would be Secondary Options. As a second example, I could specify that, as first preference, my ideal female date should have green eyes (the primary option); I could also specify that, as second preference, I would like to meet a lady with black eyes (a secondary option). It is logical to infer that my truly 'ideal' date cannot have both green and black eyes, as this becomes a physical impossibility. It is a normal occurrence in human behaviour to have preferences as reflected in the options and intensities already described. The intensity of a primary option is a primary intensity. The intensity of a secondary option is a secondary intensity. The general classification of options in primary and secondary options and the use of primary and secondary intensities are necessary if a rigorous compatibility selection/matching system is the objective.

Sumproducts - For the purpose of this invention, I will define a sumproduct as a method of combining two different strings of digits to measure the similarities between the two strings as illustrated in FIG. 4. A sumproduct for two strings is the product of an option's intensity value in the first string times its corresponding option's intensity value in the second string followed by the sum of a11 similar products obtained between the two strings. The resulting total sum of this sumproduct I call the 'score'. The score in FIG. 4 is 209 corresponding to the sumproduct between a Model and an Object (Model(x)Object). This score represents the 'full' measure of compatibility between the two strings because it acknowledges all primary and secondary intensities. A 'partial' measure would acknowledge only the Model's primary intensities and only the Object's primary intensities. An auto-sumproduct is a method of combining a string of digits with itself as illustrated in FIG. 5 in which the auto-sumproduct of the first string (the Model) produces a

scone of 255, i.e. (Model(x)Model). The Score can be described as the result of the Sumproduct but in actual use the terms can be considered practically equivalent to each other. The simplicity and usefulness of sumproducts will be apparent when I reproduce the computer results with spreadsheet calculations later in this description.

Compatibility Factor - My term 'Compatibility Factor' (CF) refers to the measurement of the degree of compatibility, percentage wise, between a selected object and the Model. The formula for CF computations is illustrated in FIG. 6. The Compatibility Factor is the ratio, in percent, of the sumproduct over the auto-sumproduct. The sumproduct is between the selected object and the Model as the example illustrated in FIG. 4. The auto-sumproduct of the Model is the reference level as the example illustrated in FIG. 5. The CF between the Object and the Model of FIG. 4 is therefore: CF = (209/255) x 100 = 81.96 For this 'full' measure of compatibility (with the Object and Subject incorporating secondary intensities) the perfect match corresponds to CF=100%. The CF can be higher than 100% if the Object string has corresponding higher intensities than the Model string, but the perfect match would still be 1000. Depending on the particular application, the Object's string can have secondary intensities or not. If the Object has secondaries then a CF of 1000 or higher is possible. If the Object's string does not have secondaries then CF=100% is hardly possible since the Object(x)Model sumproduct will probably never equal the Model(x)Model sumproduct.

By rare coincidence it could be possible to obtain a Score for a CF=100% without being a perfect match but this likelihood, specially if Objects contain no secondary intensities, is most unlikely. With millions of mathematical operations involved in each selection, Scores and CF's can become very difficult to

anticipate. Scores and CF's are to be used as quality control tools by the system's operators and not to be routinely presented, on result sheets, as it would be difficult for the average consumer to understand. In general practice, service providers and developers will find Scores and CF's not as absolute solutions but as approximate, reliable tools with the consumer being the ultimate judge of the subjective compatibility results.

Scalar - I use the term 'scalar' as the equivalent to 'weight' as used in statistical operations to modify outcomes according to relative importance. I use 'scalar' and not 'weight' to prevent confusion with the weight of people, weight of goods, etc.

Selection Engines - The Second Edition of the book "The Internet Made Simple" by P.K. McBride, ISBN 0-7506-4576-8, has a chapter listing popular 'search' engines and their techniques. The result of the searches is called 'pages' or 'hits'. The search engine databases normally contain millions of pages of information. A simple enquiry could result in many thousands of hits. By trail and error stages, the consumer can narrow down the search to meaningful answers. This takes time, ability and patience since the answers are not always direct answers. In a search engine there is no much computer processing (number crunching) as the search is mainly done by a process of elimination using logical and/or mathematical operators.

In a 'selection' engine/system all the objects in the database are processed to select the best results. That is to say, there is an essential 'number crunching' operation to implement the selection criteria. Consequently, selection databases usually do not contain millions of objects, since processing a11 the objects could become prohibitively expensive. Selection databases are normally smaller than search engine databases. But since computer power becomes cheaper every year, and selection

techniques improve, there is the possibility that selection databases could also contain millions of objects in the near future. However, to select a house, or a date, or a job, or a used car in Los Angeles does not require interrogating a database with objects also in New York, Miami or London. In commercial applications, it makes logistic sense to have smaller object databases for separate areas rather than a single large object database for the entire United States or world. The old military strategy of 'divide and conquer' applies well to selection systems. To obtain economical selection results, a normally huge selection database can be divided into several smaller selection databases (i.e. dividing books or patents into subjects or fields).

Website abbreviations - In several instances I will direct the reader to surf an Internet website to illustrate a point, as a way of example, etc. For the sake of brevity, to open a website like 'http://www.amazon.com' I will simply refer to the website by its abbreviation 'www.amazon.com'.

Example 1 - Selection of Personal Computers (PC's) A specific embodiment of the invention will now be described by way of example with reference to the accompanying drawings. In Example 1 I will present a sample selection result sheet of PC's followed by a series of controlled selection runs that gradually illustrate how the sample selection was obtained. In this example of one-way selections, a consumer searches for a PC amongst 506 computers stored in an object database. The descriptions of the computers were obtained from public sources like London newspaper adverts, magazines, catalogues, etc.

FIG. 7A and 7B shows a sample of a computer selection printout. The general appearance of this printout is common to both one-way and two-way selections. To the printout I have

inserted, in bold Italics, headings to divide up the printout into four sections as follows: Section (1) contains the particulars of the organisation providing the selection service.

Section (2) contains the particulars of the consumer (Subject) requesting the selection. In this case the prospective PC buyer has the code number of B0001500 which I will abbreviate to B- 1500.

Section (3) contains the selection results listed in five columns as follows: Column one - Contains the labels of CHOICES, CODE NUMBER, SELECTION SCORE and COMPATIBILITY FACTOR followed by 19 PC characteristics extending from FIG. 7A into FIG. 7B.

Column two - Contains the primary options of YOUR IDEAL CHOICE - the ideal PC or Model as described by the buyer in his questionnaire. Next to the SELECTION SCORE label there is the number 530540. The number 530540 corresponds to the auto- sumproduct of the Model; this number 530540 is equivalent to the perfect match between the Model and an Object containing a11 corresponding primary and secondary intensities as the Model. The Compatibility Factor of 100% corresponds, naturally, to the perfect match situation. The rest of Column two lists a11 the primary options, or first choices for each characteristic, as entered by the buyer (Subject) in his own questionnaire; because of space constrains it is not possible to list as well a11 the buyer's secondary options.

Column three - Contains CHOICE-1 which is the most compatible PC with the buyer's ideal choice. In this case, the most compatible PC available for sale is seller's code number S0000491 (Object) which I will abbreviate to S-491. The Object database has only primary intensities while the Model database has primary and secondary intensities. The SELECTION SCORE is 133110 which is the highest Model(x)Object score in the

databases and implies that, considering a11 intensities present, S-491 is 25.09% compatible with the Model as follows: (133110/530540)x100=25.09% The COMPATIBILITY FACTOR (CF) is therefore 25.09% which is the measure of compatibility. Listed below the 25.09% are the 19 primary options of Object S-491. Please note that of the total of 19 characteristics listed, S-491 has 9 primary options that match 9 primary options of the Model of Column two. Doing a hand calculation we find that (9/19)x100=47% which is a measure of compatibility that acknowledges only primary intensities in both Model and Object. The low Score and low CF of S-491 reflects the buyer's request for a powerful PC with a low price which is not a very realistic request - more about this later. I must point out that with 10 mismatches out of 19 'key words' most of the commercial search engines would have given the familiar message of "no matches available".

Column four and Column five - Contain CHOICE-2 and CHOICE-3 which are the PC's in second and third order of compatibility in accordance with the descending scores obtained. Note that the CF's also follow a descending pattern as the CF's reflect the selection scores.

At the bottom of Section (3) I list the label 'Cost of SELECTION IDENTIFICATION'. Under each PC selected from the sellers' database there is listed the payment requested (i.e. $25) for the release of the identity of each PC selected, using as reference the code numbers. Of course, this payment can be set to zero by the service provider indicating his willingness to release the results free of charge.

Section (4) contains the 'Sellers' particulars'. The 'IDENTIFICATION COUPON for B0001500' lists, under each selected PC and their code numbers, the particulars of the sellers and of the PC models selected including price. For CHOICE-1 the selected PC S-491 is a TINY Computers Home Studio 500 for $l899

(the UK price is 1186.75 assuming an exchange rate of 1=$1.60) Maximum number of selections The number of PC's selected in FIG. 7A and 7B is three but my system can select up to nine PC's. Selecting more than three PC's requires scrolling the computer screen to view results for CHOICE-4 to CHOICE-9.

General Intensity Questionnaire for PC's FIG. 8A, 8B and 8C illustrates my general questionnaire for the selection of personal computers. After minor adjustments this questionnaire will produce two suitable questionnaires for PC buyers and PC sellers respectively. FIG. 8A contains names, addresses and brand details of buyers or sellers. FIG. 88 and 8C contains the PC characteristics considered, the options for each characteristic, and clear instructions on how to enter the preferred intensities. The options made available to PC buyers and PC sellers must be identical. The options in the characteristic ACCESSORIES in FIG. 8C have only the choice of YES or NO (9 or 0 intensities) since, in this particular application, it is not practical to consider different models for each accessory.

Completed Buyer's Intensity Questionnaire for B-1500 (Subject) FIGS. 9A, 9B and 9C illustrate the completed questionnaire for buyer B-1500 (the Model). FIG. 9A contains particulars of the buyer (see Section (2) above) and I have entered a note that presupposes that the buyer insists on a PC in the $1500 to $1999 range. FIG. 9B and 9C contain all the characteristics and options considered with their respective primary or secondary intensities as entered by the buyer describing his preferences or Model. Any intensities left blank are interpreted as '0'.

These intensities are stored as a 'string' of digits in the 'subject' database.

Completed Seller's Intensity Questionnaire for S-378(Object) FIGS. 10A, 10B and 10C illustrate the completed questionnaire for seller S-378 (the object). FIG. 10A contains the particulars of the seller and the identity of the PC model for sale and its price. FIG. 10B and 10C contain all the characteristics and options considered with their appropriate primary intensities, as entered by the seller, describing the PC model for sale; for obvious reasons, secondary intensities are not present. Any intensities left blank are interpreted as '0'. These intensities are stored as a 'string' of digits in the 'object' database. Alternative "PC Buyer's Preference Questionnaire" An alternative version to my 'Intensity Questionnaires' are my 'Preference Questionnaires' in which the first preference is indicated by digit '1' and the last preference is indicated by digit '9'. The simple translation formula Intensity = (10-Preference) is applied to arrive to my software's correct intensity values, with a resultant range of intensity values between '9' and '1', without using the '0' intensity value. In practice, the intensity values of '0' or '1' have almost the same contribution to the selection process. Some consumers might prefer to use Preference Questionnaires rather than Intensity Questionnaires. In Appendix 5 I illustrate a blank "PC Buyer's Preference Questionnaire". Controlled selection Runs A, B and C To demonstrate the effect of using one-way selections, scalars and secondary intensities it is necessary to do controlled selection runs, starting with selections that use only primary

intensities. I am going to demonstrate my one-way selection system with three controlled selection runs using the following approaches: Run A - Partial compatibility (primaries only), no scalars. Run B - Partial compatibility (primaries only), with scalars.

Run B - Full compatibility (primaries and secondaries), with scalars.

Controlled selection Run A FIGS. 11A and 11B illustrate the controlled selection Run A for buyer B-1500; the questionnaire for buyer B-1500 was shown in FIGS. 9A-9C. In B-1500 I have used primaries only (eliminating secondary intensities) and no scalars. This necessitated making a copy of record B-1500, changing all intensities other than 9 to 0, and re-labelling the buyer to record B-9500.

On FIG. 11A I will focus on CHOICE-1, seller S-378 with a high CF of 73.68% because there are 14 good matches out of 19 (the questionnaire for seller S-378 was shown in FIGS. l0A-lOC). Figures 12A, 12B and 12C are a Microsoft's Excel spreadsheet illustrating the sumproduct between B-9500 and S-378, using primary intensities only. FIG. 12C shows that the selection score for S-378 is 1134. It is logical to infer that the auto- sumproduct for B-9500 (Model(x)Model) would be the number of characteristics times the primary intensities multiplied by themselves or: (19x9x9)=1539; this score of 1539 corresponds to a perfect partial match (primaries only) or CF=100%. It is also logical to infer that the Compatibility Factor between B-9500 and S-378 would be: CF=(1134/1539)x100=73.68% These results duplicate the computer results shown in FIG. 11A under SELECTIONS SCORE and COMPATIBILITY FACTOR corresponding to

'Column-two' and 'Column-three'. It is reasonable to conclude that this selection would not satisfy the customer because the price range of the PC selected S-378 is between $3500 to $ 3999 (see arrow) - this is completely out of the requested price range of $1500 to $1999; similar conclusions can be made about CHOICE-2 and CHOICE-3.

Note in FIG. 12C the abnormally high contribution of the accessories' sumproducts to the Total score. To solve these problems, scalars will be introduced.

Controlled selection Run B FIGS. 13A and 13B illustrate the controlled selection Run B for buyer B-1500. I have used primaries only, eliminating secondary intensities, and applied scalars. This necessitated making a copy of record B-1500, changing all intensities other than 9 to 0, and re-labelling the buyer to record B-9500.

On FIG. 13A I will focus on CHOICE-1, seller S-450, with 9 good matches out of 19. The score is now 78570 and the CF is now 55.43% since the sumproducts have been affected by the scalars. FIG. 14 lists the scalars used in the selection of PC's; to maintain a quantity unaffected by scalars it is the usual practice to scale the quantity by unity (1) but I will modify such quantities with a scalar of 100; the relative relationship amongst intensity products will not be affected; this will allow me to use scalars less than 100 (my new unity) without introducing decimal points; the arbitrary criteria I used in selecting the scalars for PC's was to make the scalars approximately equal to the factory price of the PC components in US Dollars.

FIGS. 15A, 15B and 15C are a Microsoft's Excel spreadsheet illustrating the sumproduct between B-9500 and S-450 (Model(x)Object), using primary intensities only and applying the scalars listed in FIG. 14. FIG. 15C shows that the

selection score for S-450 is 78570 which duplicates the computer score in FIG. 13A, Column-three.

In FIGS. 16A, 16B and 16C I show the spreadsheet for the auto- sumproduct with scalars of the model B-9500 (Model(X)Model) with a total score of 141750; this score duplicates the SELECTION SCORE of 'Column-two' in FIG. 13A.

I can now compute by hand the CF as follows: CF=(785701141750)x100=55.430; this duplicates the CF of S-450 as shown in 'Column-three' on FIG. 13A.

Note that the APPROXIMATE PRICE of S-450 is now '$1500 to $1999' (see arrow) and hence a good match with the model, as the consumer insisted. Similar observation can be made about CHOICE-3. These improved results are on account of the high scalar of 500 used for the APPROXIMATE PRICE characteristic. Note in FIG. 15C the nominal contribution of the accessories' sumproducts. The CF for CHOICE-1 of 73.68% in Run A has decreased to 55.43% in Run B - this is the trade off for a more balanced and improved overall compatibility. Note that S- 378 was CHOICE-1 in Run A and now has been relegated to CHOICE-2 in Run B.

Controlled selection Run C FIGS. 17A, 17B, 17C and 17D illustrate the controlled selection Run C for buyer B-1500. I have used primary intensities for the Objects and primary and secondary intensities for the Model, and applied the scalars of FIG. 14.

On FIG. 17A I will focus on CHOICE-1, seller S-491, with 9 good matches out of 19. The score is 133110 and the CF is now 25.09% since the sumproducts have been affected by the scalars.

FIGS. 18A, 18B and 18C are a Microsoft's Excel spreadsheet illustrating the sumproduct between B-1500 and S-491

(Model(x)Object), using primary and secondary intensities, and applying the scalars listed in FIG. 14.

FIG. 18C shows that the selection score for S-491 is 133110 which duplicates the selection score in FIG. 17A, Column-three. In FIGS. 19A, 19B and 19C I show the spreadsheet for the auto- sumproduct of the Model B-1500 (Model(x)Model) with a total score of 530540 (the maximum possible score using both primary and secondary intensities); this score duplicates the SELECTION SCORE of the model in Column-two in FIG. 17A.

I can now manually compute the CF as follows: CF=(133110/530540)x100=25.09%; this CF duplicates the CF of S-491 as shown in Column-three on FIG. 17A.

Note that the APPROXIMATE PRICE of S-491 is '$1500 to $1999' (see arrow), and hence a good match with the model, as the consumer insisted. Similar observation can be made about CHOICE- 3.

In Run B, CHOICE-1, the CF of 55.43% has decreased to 25.09% in Run C - this low CF indicates that, considering a11 intensities, either the object database is not large enough or that the consumer requirements are very stringent or both.

Note that S-450 was CHOICE-1 in Run B and has now been relegated to CHOICE-3 in Run C.

FIGS. 17C and 17D contain the remaining six PC's selected as CHOICE-4 to CHOICE-9 that are viewed by scrolling the monitor screen (I did not consider it necessary to reproduce their Identification Coupons). The nine PC's selected are the most compatible in the database with the Model specified - the final PC selected by the consumer will ultimately be a subjective decision and my system provides information that the consumer can use to help him make such decision.

It is reasonable to extrapolate that, having a larger Object database and sensible consumer requirements, my system would

deliver much higher values of CF; and that, having a perfect match present in the Object database, my system will positively identify the perfect match, as I will demonstrate later in this description.

In this PC application of Run C the CF's can probably never reach 100% since the PC Objects do not contain secondary intensities while the PC Model do contain secondary intensities - the Object(x)Model sumproducts will probably never equal the Model's auto-sumproduct Model(x)Model; in other applications, like computer dating, some of the Object characteristics do contain secondary intensities. My system's software is flexible and can always be modified to accommodate other different methods of computing CF's. Run C of FIGS. 17A and 17B is also the sample result sheet initially presented as FIGS. 7A and 7B.

Compatibility Comparison - Summary of Results for B-1500 FIG. 20 shows, in summary form, the progression of the compatibility results from Run A to Run B to Run C. In Run A CHOICE-1, or first place, was S-378. In Run B, S-378 was relegated to second place with S-450 as newly selected on first place. In Run C, S-450 was relegated to third place with S-491 as newly selected on first place.

I maintain that, in going from Run A to Run C, there has been a major improvement in the quality of the overall compatibility selection results due to the introduction of scalars and secondary intensities. But, as I have stated before, in the highly subjective matter of selections, it is up to the consumer to accept or to reject the selections of Run C.

Correctly speaking, we can not compare the CF's of Runs A, B and C against each other as the CF's were obtained under different conditions; we can compare the CF's against each other only within each specific run (i.e. within Run A, or Run B or Run C) and such comparison is satisfactory.

Integrity Selection Runs I will present below several selection runs designed to demonstrate the integrity and versatility of my system as well as to consolidate some of the concepts already discussed The Perfect Match with CF--100% for B-9491 In Run C, FIG. 17A, for buyer B-1500 CHOICE-1 was seller S-491 with CF=25.090. By creating a new buyer with characteristics exactly identical to seller S-491 then we should get a perfect match. This necessitated taking record S-491 from the sellers' database, making a copy of it, moving copy S-491 to the buyers' database and re-labelling the new buyer as B-9491 since this new buyer has only primary and no secondary intensities.

FIGS. 21A, 21B, 21C, 21D, 21E and 21F show the selection run for B-9491. FIG. 21A, shows that indeed S-491 has been selected as CHOICE-1 for B-9491; the seller's Score for S-491 of 141750 and CF=100% exactly duplicate the Model's Score and CF of Column two. The buyer's and seller's Scores are both 141750 because the sumproduct Model(x)Object equals the auto-sumproduct Model(x)Model, hence CF=100%. The list of all the 19 characteristics of seller S-491 have a perfect match with the all the primary characteristics of buyer B-9491. FIG. 21B lists again S-491 as the TINY Computers System Home Studio 500 for $1899.

FIGS. 21C and 21D shows all the 9 computers selected, all within the price range requested of $1500 to $1999. FIGS. 21E and 21F show the IDENTIFICATION COUPON for B-9491 with the names of the manufacturers of the 9 PC models with their occurrences as follows: TINY=2 models, COMPAQ=3 models, TIME=1 model, DELL=2 models and VIGLEN=1 model. This means that, for the PC price range of $1500 to $1999, the TINY model S-491 has a well- balanced competition. Should a11 or most of the 9 PC's selected

peen TINY models then this could have meant that part of the TINY PC product range was under-priced, overpriced or repetitive, an important consideration for the TINY marketing department. This 'perfect match' exercise indicates that my system can also be used as a marketing diagnosis tool. The Perfect Match with CF=34.77% for B-491 In Run C, FIG. 17A, for buyer B-1500 CHOICE-1 was seller S-491 with CF=25.09%. By creating a new buyer with primary characteristics exactly identical to seller S-491 then we should get a perfect match. This necessitated taking record S-491 from the sellers' database, making a copy of it, moving copy S-491 to the buyers' database and re-labelling the new buyer as B-491. Since the copy had only primary and no secondary intensities, I. added suitable secondary intensities to reproduce the configuration of a normal PC buyer's record.

FIGS. 22A and 22B show the selection run for buyer B-491. FIG. 22A shows that indeed S-491 has been selected as CHOICE-1 for B-491 with a seller's Score (Model(x)Object) of 141750. The Model's Score (Model(x)Model) of B-491 in Column two is now a high 407730 because of the extra contribution of the added secondaries. These two values of Scores produce a CF=34.77%. The list of all 19 characteristics of seller S-491 have a perfect match with all the primary characteristics of the Model of buyer B-491. The resulting value of CF clearly depends on the configuration of the reference level (Model(x)Model) selected. FIG. 22B lists again S-491 as the TINY Computers System Home Studio 500 for $1899. Reverse Selection for Seller S-491 In Run C, FIG. 17A, the selection for buyer B-1500 CHOICE-1 was seller S-491 with a Score of 113110 and CF=25.09%; the Model's score for B-1500 in Column-two was 530540. If a

'reverse' selection was to be run selecting buyers for seller S- 491 (buyers now taking the Object role), then my system should be able to select buyer B-1500 as one of the Objects selected.

FIG. 23 shows a selection run for seller S-491 (to fit the page the SELECTER's Tel. and Fax. lines have been removed) in which seller S-491 has taken the role of a Subject; Column-five, CHOICE-3, indicates that buyer B-1500 has been selected as an Object as anticipated; the Score is 133110 which duplicates the Score for seller S-491 in FIG. 17A, Column-three. The new CF in FIG. 23, Column-five is CF=93.90o because this time the Model's auto-sumproduct of Column-two has a low Score of 141750 since S- 491 contains no secondary intensities.

Please remember that I created two new buyers with identical primary characteristics as seller S-491; they were buyer B-491 with primary and secondary intensities, and B-9491 with only primary intensities. In FIG. 23, B-491 has been selected as CHOICE-1 (Column-three) and B-9491 as CHOICE-2 (Column-four); for obvious reasons, the Score and CF of both buyers B-491 and B-9491 are 141750 and 100% respectively. This 'reverse selection' exercise could also be used as a marketing diagnosis tool. Influencing Selection Outcomes Using Scalars In Run C, FIG. 17A, Column-two, buyer B-1500 has requested a powerful PC system for the low price range of $1500 to $1999. After seeing the results of Run C, the buyer realises that his preparedness to pay a low-price for his system is not a realistic one and wishes to modify his specifications. The buyer now insists on approximately the same system originally requested, at a more realistic price, but with the proviso that he must get a 19-inch monitor.

I can influence the selection outcome by changing the values of the scalars listed in FIG. 14. 1 changed the MONITOR scalar

from 150 to 550 and changed the APPROXIMATE PRICE scalar from 500 to zero. FIG. 24 shows the selection results for buyer B- 1500 with the modified scalars. All the PC systems selected are powerful computers with characteristics more compatible with the Model than in FIG.17A. Note that all systems selected have 19- inch monitors because the change in scalar for monitors from 150 to 550 greatly increased the relative importance of the characteristic MONITOR.

The APPROXIMATE PRICE of the PC systems selected are over $3000 reflecting the high power of the systems; by changing the price scalars from 500 to zero the intensity values associated with the APPROXIMATE PRICE characteristic have no influence whatsoever on the selection Score even though the APPROXIMATE PRICE listed for buyer B-1500's Model still appears as '$1500 to $1999' as entered in the buyer's questionnaire. In Internet applications, the scalars listed in FIG. 14 can be presented to the consumer as default scalar values and, by extending to the consumer the ability to modify the default scalars a great flexibility to the selection process can be obtained. Example 2 - Selection of Social Partners (Computer Dating) A specific embodiment of the invention will now be described by way of example with reference to the accompanying drawings. In Example 2 I will present a sample selection result sheet of computer dating followed by a series of controlled selection runs that gradually illustrate how the sample selection was obtained.

In this example of 'two-way' or 'reciprocal' selections, a male searches for a compatible female partner. There are 243 females stored in an 'object' database; the female database contains characteristics for the actual females and their ideal male partners. There are also 243 males stored in a 'subject'

database; the male database contains characteristics for the actual males and their ideal female partners. The descriptions of the males and females were fabricated; their fictitious locations are close to the London subway network.

FIG. 25A and 25B show a sample of a computer dating printout. The general appearance of this printout is common to both one- way and two-way selections. To the printout I have inserted, in bold Italics, headings to divide up the printout into four sections as follows: Section (1) contains the particulars of the organisation providing the selection service.

Section (2) contains the particulars of the male (Subject) requesting the selection. In this case the prospective male partner has the code number of M0000015 which I will abbreviate to M-15.

Section (3) contains the selection results listed in six columns as follows: Column one - Contains the labels of CHOICES, CODE NUMBER, SELECTION SCORE and COMPATIBILITY FACTOR followed by 17 personal characteristics extending from FIG. 25A into FIG. 25B.

Column two - Contains the primary options of YOURSELF, in this case, the male M-15 requesting the female partner.

Column three - Contains the primary options of YOUR IDEAL CHOICE - the ideal female or Model as described by the male in his questionnaire. Next to the SELECTION SCORE label there is the number 1716500. The number 1716500 corresponds to the auto- sumproduct of the Model; this number 1716500 is equivalent to the perfect match between the Model and an Object containing a11 corresponding primary and secondary intensities as the Model. The Compatibility Factor of 100% corresponds, naturally, to the perfect match situation. The rest of Column three lists all the primary options, or first choices for each characteristic, as entered by the male (Subject) in his own questionnaire; because

of space constrains it is not possible to list as well a11 the male's secondary options.

Column four - Contains CHOICE-1 which is the most compatible female with the male's ideal choice. In this case, the most compatible female available for dating is female's code number F0000011 (Object) which I will abbreviate to F-11. The Object database and the Model database they both have primary and secondary intensities. The SELECTION SCORE is 775150 which is the highest Model(x)Object score in the databases and implies that, considering all intensities present, F-11 is 45.16% compatible with the Model as follows: (775150/1716500)x100=45.160 The COMPATIBILITY FACTOR (CF) is therefore 45.16% which is the measure of compatibility listed. Listed below the 45.16% are the 17 primary options of Object F-11. Please note that of the total of 17 characteristics listed, F-11 has 7 primary options that match 7 primary options of the Model of Column three. Doing a hand calculation we find that (7/17)x100=41.18% which is a measure of compatibility that acknowledges only primary intensities in both Model's and Object's own descriptions. I must point out that with 10 mismatches out of 17 'key words' most of the commercial search engines would have given the familiar message of "no matches available".

Column five and Column six - Contain CHOICE-2 and CHOICE-3, which are the females in second and third order of compatibility, in accordance with the descending scores obtained. Note that the CF's also follow a descending pattern as the CF's reflect the selection scores.

At the bottom of Section (3) I list the label 'Cost of SELECTION IDENTIFICATION'. Under each female selected from the females' database there is listed the payment requested (i.e. $100) for the release of the identity of each female selected, using as reference the code numbers. Of course, this payment can

be set to zero by the service provider indicating his willingness to release the results free of charge.

Section (4) contains the females' particulars. The 'IDENTIFICATION COUPON for M-15' lists, under each selected female and their code numbers, the particulars of the three females selected. For CHOICE-1 the selected female F-11 is Miss Kathy Lady-11 of Queensbury, London. Maximum number of selections The number of social partners selected in FIG. 25A and 25B is three but my system can select up to eight partners. Selecting more than three partners requires scrolling the computer screen to view results for CHOICE-4 to CHOICE-8.

General 'Computer Dating - Intensity Questionnaire' FIGS. 26A, 26B, 26C, 26D and 26E illustrate my general intensity questionnaire for the selection of social partners. The same questionnaire is completed by either male or female clients as the options and required intensities are identical for both male and female clients.

FIG. 26A contains the name and address of the client. The remainder of the questionnaire is divided into two parts; the first part contains characteristics about the client itself and the second part contains characteristics about the client's ideal partner (the Model); both of these two parts contain identical sets of characteristics and options.

FIGS. 26B and 26C contain the characteristics considered for the client itself, the options for each characteristic, and clear instructions on how to enter the preferred intensities. For obvious reasons most of FIG. 26B requires only primary intensities and most of FIG. 26C requires primary and secondary intensities.

FIGS. 26D and 26E contain the characteristics considered for

the client's ideal partner (the Model), the options for each characteristic, and clear instructions on how to enter the preferred intensities. For obvious reasons all of FIGS. 26D and 26E require primary and secondary intensities.

Completed Male's Intensity Questionnaire for M-15 (Subject) FIGS. 27A, 27B, 27C, 27D and 27E illustrate the completed questionnaire for male M-15. FIG. 27A contains particulars of the male client (see Section (2) above).

FIGS. 27B and 27C contain a11 the characteristics and options considered, with their respective primary and secondary intensities, as entered by the male client describing himself. Any intensities left blank are interpreted as '0'. These intensities are stored as a 'sub-string' of digits in the 'subject' database.

FIGS. 27D and 27E contain a11 the characteristics and options considered, with their respective primary and secondary intensities, as entered by the male client describing his ideal female partner (his Model). These intensities are stored as a second 'sub-string' of digits in the 'subject' database.

Completed Female's Intensity Questionnaire for F-10 (Object) FIGS. 28A, 28B, 28C, 28D and 28E illustrate the completed questionnaire for female F-10. FIG. 28A contains particulars of the female client (see Section (2) above).

FIGS. 28B and 28C contain a11 the characteristics and options considered, with their respective primary and secondary intensities, as entered by the female client describing herself. Any intensities left blank are interpreted as '0'. These intensities are stored as a 'sub-string' of digits in the 'object' database.

FIGS. 28D and 28E contain a11 the characteristics and options considered, with their respective primary and secondary

_ntensities, as entered by the female client describing her ideal male partner (her Model). These intensities are stored as a second 'sub-string' of digits in the 'object' database. I will discuss the handling of these 'sub-strings' later under 'The Master String'. Alternative "Computer Dating Preference Questionnaire" An alternative version to my 'Intensity Questionnaires' are my 'Preference Questionnaires' in which the first preference is indicated by digit '1' and the last preference is indicated by digit '9'. The simple translation formula Intensity = (10-Preference) is applied to arrive to my software's correct intensity values, with a resultant range of intensity values between '9' and '1', without using the '0' intensity value. In practice, the intensity values of '0' or '1' have almost the same contribution to the selection process. Some consumers might prefer to use Preference Questionnaires rather than Intensity Questionnaires. In Appendix 6 I illustrate a blank "Computer Dating Preference Questionnaire". Controlled selection Runs D, E, and F To demonstrate the effect of using two-way (reciprocal) selections, scalars and secondary intensities, it is necessary to do controlled selection runs, starting with selections that use only primary intensities. I am going to demonstrate my two- way s-lection system with three controlled selection runs using the following approaches: Run D - Partial compatibility (primaries only), no scalars. Run E - Partial compatibility (primaries only), with scalars.

Run F - Full compatibility (primaries and secondaries), with scalars.

Controlled selection Run D FIGS. 29A and 29B illustrate the controlled selection Run D for Male M-15 in which secondary intensities have been eliminated; the questionnaire for male M-15 was shown in FIGS. 27A-27E. In M-15 I have used primaries only and no scalars. This necessitated making a copy of record M-15, changing a11 intensities other than 9 to 0, and re-labelling the buyer to record M-9015. All the other male records were treated equally as well, with the result that the male database contains now records M-1 to M-243 and M-9001 to M-9243.

To prepare the databases for the reciprocal interrogation of primaries only, all the female records were treated equally as well, with the result that the female database contains now records F-1 to F-243 and F-9001 to F-9243.

On FIG. 29A I will focus on CHOICE-1, female F-9010 with a CF of 61.760, with 10 good matches out of 17 (the questionnaire for female F-9010 was shown in FIGS. 28A-28E).

Figures 30A, 30B, 30C, 30D, 30E and 30F are a Microsoft's Excel spreadsheet illustrating the sumproduct between M-9015 and F-9010, using primary intensities only. FIG. 30F shows that the spreadsheet selection score for F-9010 is 1701.

It is logical to infer that the auto-sumproduct for M-9015 (Model(x)Model) would be the number of characteristics times the primary intensities multiplied by themselves or: ((17+17)x9x9))=2754; this score of 2754 corresponds to a perfect partial match (primaries only) or CF=100o.

It is also logical to infer that the Compatibility Factor between M-9015 and F-9010 would be: CF=(1701/2754)x100=61.760 This result duplicate the computer results shown in FIG. 29A under SELECTIONS SCORE and COMPATIBILITY FACTOR corresponding to 'Column-three' and 'Column-four'. Please note that the CF reference level is related to 34 characteristics; the 10 good

matches in FIG. 29A is related to the 17 characteristics described in FIGS. 30D-30F corresponding to the matches between the actual female and the female model; the other 17 matches between the actual male M-9015 and F-9010's male model, as seen on FIGS. 30A-30C, are not displayed on the result sheets.

It is reasonable to assume that this selection of F-9010 would not satisfy the Jewish male client M-9015 because he wishes to meet a Jewish lady and F-9010 is a Moslem lady, a serious mismatch as indicated by the arrows on FIG. 29A; similar assumptions can be made about CHOICE-2 and CHOICE-3. To solve this problem, scalars will be introduced.

Controlled selection Run E FIGS. 31A and 31B illustrate the controlled selection Run E for male M-15. I have used primaries only, eliminating secondary intensities, and have applied scalars. This necessitated making a copy of record M-15, changing all intensities other than 9 to 0, and re-labelling the buyer to record M-9015. All the other male records were treated equally as well, with the result that the male database contains now records M-1 to M-243 and M-9001 to M-9243.

To prepare the databases for the reciprocal interrogation of only primaries, all the female records were treated equally as well, with the result that the female database contains now records F-1 to F-243 and F-9001 to F-9243.

On FIG. 31A I will focus on CHOICE-1, female F-9010, with 10 good matches out of 17. The score is now 481950 and the CF is 76.28% since the sumproducts have been affected by the scalars.

FIG. 32 lists the scalars used in the selection of social partners; to maintain a quantity unaffected by scalars it is the usual practice to scale the quantity by unity (1) but I have modified such quantities with a scalar of 100; the relative relationship amongst intensity products will not be affected;

this will allow me to use scalars less than 100 (my new unity) without introducing decimal points; I have used an arbitrary criteria in selecting the scalars for social partners.

FIGS. 33A, 33B, 33C, 33D, 33E and 33F are a Microsoft's Excel spreadsheet illustrating the sumproduct between M-9015 and F9010 (Model(x)Object), using primary intensities only and applying the scalars listed in FIG. 32.

FIG. 33F shows that the selection score for F-9010 is 481950 which duplicates the computer score in FIG. 31A, Column-four. In FIGS. 34A-34F I show the spreadsheet for the auto- sumproduct, with scalars, of the model M-9015 (Model(X)Model) with a total score of 631800; this score duplicates the SELECTION SCORE of 'Column-three' in FIG. 31A.

I can now compute by hand the CF as follows: CF=(481950/631800)x100=76.28%; this duplicates the CF of F-9010 of 76.28% as shown in 'Column- four' on FIG. 31A.

Note that the religion of CHOICE-1 is still Moslem but now a Jewish lady appears as CHOICE-2 (see arrows) and hence a set of improved selections is beginning to appear. These improved results are on account of the scalar of 200 used for the RELIGION characteristic.

Controlled selection Run F FIGS. 35A, 35B, 35C and 35D illustrate the controlled selection Run F for male M-15. I have used primary and secondary intensities for the Objects and for the Model, and applied the scalars of FIG. 32.

On FIG. 35A I will focus on CHOICE-1, female F-11, with 7 good matches out of 17. The score is 775150 and the CF is now 45.16% since the sumproducts have been affected by the scalars.

FIGS. 36A, 36B, 36C, 36D, 36E and 36F are a Microsoft's Excel spreadsheet illustrating the sumproduct between male M-15 and

female F-11 (Model(x)Object), using primary and secondary intensities, and applying the scalars listed in FIG. 32. FIG. 36F shows that the selection score for F-11 is 775150 which duplicates the computer score in FIG. 35A, Column-four.

In FIGS. 37A, 37B, 37C, 37D, 37E and 37F I show the spreadsheet for the auto-sumproduct of the Model M-15 (Model(x)Model) with a total score of 1716500 (the maximum possible score using both primary and secondary intensities); this score duplicates the SELECTION SCORE of the model in Column-three in FIG. 35A.

I can now manually compute the CF using the scores from the spreadsheets as follows: CF=(775150/1716500)x100=45.16%; this CF duplicates the CF of F-11 as shown in Column-four on FIG. 35A.

Note that the religion of CHOICE-1 is now of a Jewish lady, and that a second Jewish lady appears as Coice-3 (see arrows). Note that F-10 was CHOICE-1 in Runs D and E and has now been relegated to CHOICE-2 in Run F.

FIGS. 35C and 35D contain the remaining five females selected as CHOICE-4 to CHOICE-8 that are viewed by scrolling the monitor screen. The eight females selected are the most compatible in the database with the Model specified - the final female selected by the consumer will ultimately be a subjective decision and my system provides information that the consumer can use to help him make such decision. It is reasonable to extrapolate that, having a larger Object database and sensible consumer requirements, my system would deliver much higher values of CF; and that, having a perfect match present in the Object database, my system will positively identify the perfect match.

In this application of Run F the CF's could reach 100% since the Subjects and Objects contain exactly the same number of 171

equivalent intensities as entered on the questionnaires. Run F of FIGS. 35A and 35B is also the sample result sheet initially presented as FIGS. 25A and 25B. Compatibility Comparison - Summary of Results for M-15 FIG. 38 shows, in summary form, the progression of the compatibility results from Run D to Run E to Run F. In Run D CHOICE-1, or first place, was Moslem lady F-9010. In Run E, F- 9010 continued as CHOICE-1 but the Jewish lady F-9011 emerged as CHOICE-2. In Run F, F-9010 was relegated to second place (as F- 10) and the Jewish lady F-9011 was moved to first place (as F- 11). A second Jewish lady, F-16, emerged as CHOICE-3.

I maintain that, in going from Run D to Run F, there has been a major improvement in the quality of the overall compatibility selection results due to the introduction of scalars and secondary intensities. But, as I have stated before, in the highly subjective matter of selections, it is up to the consumer to accept or to reject the selections of Run F.

I can not compare the CF's of Runs D, E and F against each other as the CF's were obtained under different conditions; I can compare the CF's against each other only within each specific run (i.e. within Run D, or Run E or Run F) and such comparison is satisfactory.

Reverse Selections (i.e. selection of male partners for a female client) In FIGS. 39A and 39B I present the 'reverse' results of selecting male partners for female F-11 who was selected as the most compatible partner for male M-15 in the 'forward' Run F, FIG. 35A.

In FIG. 39A, CHOICE-1 for F-11 is male M-15, a 'reverse' reproduction of the 'forward' results of Run F; this is another example of the integrity of my system. But there is no guarantee

that the two most compatible partners in a forward selection will also be reproduced as the two most compatible partners in the reverse selection - the 'ideal' female for a male does not guarantee that the male will be the 'ideal' male for the female and very large databases would confirm this fact. The same computer program is used for the selection of forward and reverse selections.

I have constructed two dating databases: a male and a female database. The first database entered into the selection module is always treated as the Subject and the second is treated as the Object. In Run F, FIG.35A, the females were the objects selected and in FIG. 39A the males were the objects selected; this reversal of roles is accomplished by a simple reversal of the order of database entry into the selection module. Graphical Representation of The Compatibility Technique I wish to present a graphical equivalent of my compatibility technique. In FIGAO I have plotted a graph for the CHARACTERISTIC "Taste in Food". The Intensities are plotted on the vertical axis and seven food Options are plotted on the horizontal axis. The curve M represents the Subject, or Model, with intensities 6,7,8,9,8,7 and 6. The other three curves represent Objects N, P and Q with their respective intensities. From the geometry it is obvious that curve N is the closest (the most compatible) to curve M. The second closest to curve M is curve P and the third closest is curve Q.

FIG. 41 shows the sumproducts for the above four curves that confirms mathematically the decreasing compatibility, with respect to curve M, of curves N, P and Q respectively as observed on the geometry of FIG. 40. The sumproducts between two curves (i.e. M(x)N) were obtained by multiplying respective pairs of intensity values (Im x In) for each food option and summing the products.

Rare Selection Results In a11 the selections performed in this detailed description there has been no cases of unusual results. But, for the record, I must point out the remote possibility of some unusual results. In my selection of personal computers the Subjects have primary and secondary intensities while the Objects have only primary intensities. In my selection of social partners the Subjects and the Objects both have primary and secondary intensities. Whenever Objects have secondary intensities two rare selection results are possible: unusual Score results and unusual CF results.

Unusual Score Result FIG. 42 shows a situation in which the Object string has some-.. intensities of higher value that the Subject's (Model) intensities; the resulting sumproduct Score is 272. The Subject's auto-sumproduct of FIG. 43 is 255. These two sumproduct values produce a compatibility factor of 106.67% as shown in FIG. 44. This CF higher than 100% is the signal that the Object selected can not be a perfect match, but still the selection is an approximate, and very probably acceptable, compatibility result.

Unusual Compatibility Factor Result FIG. 45 shows a situation in which the Object string has intensities different from the Subject's intensities; the sumproduct score is 255. FIG. 46 shows that the Subject's auto- sumproduct is also 255. The resulting CF of FIG. 47 is therefore 100% even though the Object and Subject strings are not identical, but still very similar to each other. We must remember that a typical string has hundreds of millions of intensity permutations and that the probability of two such strings producing anomalous values of CF are extremely rare.

Still, the massive power of the 'number crunching' operation delivers an approximate, and very probably acceptable, compatibility selection result.

The TITAN Computer Program The actual implementation of the present invention is by means of an algorithm in a software computer program that is stored in, and that controls the operation of, a digital computer. For convenience I am going to refer to my software as the TITAN Computer Program and my compatibility system as the TITAN Selection System (sm).

TITAN's source code is written in FORTRAN-77 language. Fortran statements are typed in lines of up to 72 characters and the program interprets the characters according to their position (column number) on the statement line. For the typed text of my application description to be in alignment with Fortran statements, and selection results, I am using in my application the Courier-New font that uses equal character spacing.

After a questionnaire is completed, the new data collected needs to be entered into the TITAN system. As an example, I am going to illustrate entering PC seller S-378 into the TITAN system; I presented the completed intensity questionnaire of S- 378's on FIGS. l0A-10C. Computer Seller S-378 questionnaire's details are entered into a specially formatted text file called the "CompSell.txt" file. FIGS. 48A, 48B, 48C and 48D show a listing of the CompSell.txt file; I have entered in Italics additional comments to help with the explanation. On FIG. 48A the bullet .INIT initialises the building of the CompSell.txt file.

Next follows "The Master String". The Master String is a special sub-file that defines the structure of the databases built by the TITAN program and controls the sumproducts performed during the selection stage. The organisation of the

Master String duplicates the organisation of the questionnaire. The advantage of the Master String (and associated questionnaire) is that it can easily be modified to suit other applications without changing the basic source code, a situation welcomed by commercial Application Service Providers (ASP).

A block of 21 format statements continues, corresponding to the 21 lines of information contained on the first page of the seller's questionnaire of FIG. 10A; a blank line follows. The 'ONE-WAY' statement signifies that the file will be involved in one-way selection only.

Next, data is entered in groups corresponding to each CHARACTERISTC (i.e. PC CHASSIS TYPE) followed by the Options that further defines the Characteristic. A blank line signifies the end of a particular Characteristic. In this PC case,-each Option is to be recognised by a one-digit figure. FIGS. 48A-48C show a11 the 19 Characteristics of the PC seller's questionnaire.

At the middle of FIG. 48C the bullet .ADD signifies that a new seller record(s) is to be added to the CompSell.txt file. Entering seller's code number S0000378 signifies the beginning of the new record. Next follows the particulars of the seller in accordance with the format specified on FIG. 48A.

On FIG. 48D I list the "intensity string" as entered on the seller's questionnaire and corresponding to the respective 19 Characteristics. Any mistake made by entering too many, or too few, intensity values is detected by the TITAN system, giving an error message with the line number where the error occurs; this safeguards the integrity of the entire database.

The bullet .LIST indicates the intention of generating a print file "PRINT1.txt", for quality control purposes, after the database has been constructed. The bullet .END signifies the end of the operation. A similar text file, "CompBuy.txt", is built collecting the data entered on the buyer's questionnaire.

There could be any number of Characteristics in a Master String and each Characteristic could have any number of Options. The total number of Options equals the sum of all Options in a11 Characteristics. The total number of Options could be any number.

To perform sumproducts the intensity values in the 'intensity string' are positioned end-to-end. In one-way selections (i.e. PC selections) the Subject's intensity string is sumproduct in sequence against all the Objects' intensity strings. In two-way selections (i.e. computer dating) the intensity strings are divided into two sub-strings: one sub-string for the Client and another sub-string for his/her Model; the result is that one Subject and one Object have a total of four sub-strings to be cross-sumproduct in appropriate pairs.

In selecting a female date for a male client, a11 the females are cross-sumproduct against the male client; in selecting a male date for a female client, all the males are crosssumproduct against the female client; these two reverse operations could produce identical results (see FIGS. 35A and 39A) but there is no guarantee of such identical results. If in a selection run for client John, lady Doris is selected as his ideal female, this does not necessarily means that John is the ideal male for Doris since only John's particulars have been interrogated in the male database; a separate computer run is necessary in which all males are interrogated to identify the ideal male for client Doris.

The TITAN Source Code The TITAN program consists of two modules: TITAN1 and TITAN2. The TITAN1 module builds selection databases and the TITAN2 module performs the selection operations using the databases. My Fortran-77 TITAN Source Code is attached: Appendix 7 contains the source code for TITAN1 and Appendix 8 contains the source code for TITAN2. 56

TITAN1 Module The module that builds selection databases is TITANI. FIG. 49 shows a flowchart for building databases for the selection of personal computers, that is, a database for sellers and a database for buyers of PC's.

On the right top corner of FIG. 49, the data collected on the seller's questionnaire is used to build the text file CompSell.txt; we have examined in detail this file on FIGS. 48A- 48D. Next, we execute the program 'TITANI.exe' to build the sellers' database 'Database.s'. As an example, FIGS. 50A and 50B show the listing of the 'PRINTl.txt' quality control file depicting the insertion of seller S-378 (see arrows) into the sellers' database; to focus on the insertion, the listing has been condensed after removing 505 other records. FIG. 50B clearly illustrates seller S-378's intensity string.

The same TITAN1 module is used to build databases for PC buyers, male databases, female databases, etc. after simple modifications to the CompSell.txt file and its Master String. FIG. 51 shows the TITAN1 flowchart for building dating databases for male and female partners. TITAN2 Module The module that performs the selection of Objects is TITAN2. FIG. 52 shows a TITAN2 flowchart for the selection of PC's. On the left of FIG. 52 I show the buyers' database 'Database.b' and on the right the sellers' database 'Database.s'. The first of these two databases that are read in becomes the Subject and the second read in becomes the Object.

*The top of FIG. 52 shows the file 'Card2.txt' containing the Run-Time-Parameters (RTP) needed for the execution of the TITAN2 program 'TITAN2.exe'. FIG. 53 shows two versions of the 'Card2.txt' file with RTP's for the selection of PC's, with no

scalars and with scalars.

The bottom of FIG. 52 shows the file 'Print2.txt' which presents the results of the selection of PC's, such as the results of Run C, FIGS. 17A-17D.

FIG. 54 shows a TITAN2 flowchart for the selection of male and female social partners. FIG. 55 shows two versions of the 'Card2.txt' file with RTP's for the selection of social partners, with no scalars and with scalars.

The bottom of FIG. 54 shows the file 'Print2.txt' which presents the results of the selection of social partners, such as the results of Run F, FIGS. 35A-35D.

APPENDICES Appendix 1 email from J. Guerrero to Amazon Appendix 2 email from Amazon to J. Guerrero Appendix 3 sample of "3w.bestmatch.com" questionnaire Appendix 4 sample of "3w.matchmakerintl.com" questionnaire Appendix 5 blank "PC Buyer's Preference Questionnaire" Appendix 6 blank "Computer Dating Preference Questionnaire" Appendix 7 TITAN1 Fortran-77 source code Appendix 8 TITAN2 Fortran-77 source code

Read Message Help as attachment Back to Sent Next Download Attachments Choose Fo Date: Sat, 14 Aug 1999 14:22:11 -0700 (PDT) Add Addresses From: Jose Guerrero < joseguerrerolondon@yahoo.com> I Block address Subject: Jose to Amazon, math topics, 14 Aug 99 To: feedback@amazon.com ;To: feedback@amazon.com From: JoseGuerreroLondon@Yahoo.Com Date: 14 August 1999 Subject: Need mathematics book Dear Amazon, I would like to purchase several copies of a mathematics book with the basic principles of the following topics: i 1) Linear Equations 2) Exponential and Logarithmic Functions j 3) Matrix Algebra 4) Linear Programming 5) Probability Theory 6) Differential Calculus 7) Integral Calculus I have not been able to find in your search service any single book with all the above topics. I will appreciate it very much if you would guide me in how to search for a book with the above topics. Best re:..;ards, Jose Guerrero JoseGuerreroLondon@Yahoo.Com

Read Message Help as attachment 7 Back to Inbox p4 Prey I Next Download Attachments Choose Folder ;; Date: Sun, 15 Aug 1999 04:48:18 -0700 (PDT) Add Addresses From: < feedbaek@amazon.eom> I Block address To: Jose Guerrero < joseguerrerolondon@yahoo.com> Subject: Your Feedback to Amazon.com From: feedback@amazon.com Dear Jose, Greetings from Amazon.com! Unfortunately, please know that we do not currently don't have a way to search by topics or table of contents. However, I think this is a great suggestion, and I've passed it on to the appropriate department in our company for consideration. We truly value this kind of feedback, as it helps us continue to improve our store and provide better service to our customers. For your reference, the only information we can give you about particular titles is what we display on the detail page for that item. Usually, we have the title, author; edition, publisher, publication date, price, any US Library of Congress subject categories, and ISBN. Sometimes we have a short synopsis, review, or graphical image of the cover. We know that, often, the amount of information we have may make it difficult to make a purchasing decision unless you already know something about the item. Our editorial staff is working on building up the amount of information we provide about each item. We know that excerpts, reviews, and blurbs are very important to buyers, and we will continue to explore new ways to provide customers with this kind of valuable information. Otherwise, in cases such as yours we recommend that you post your question to a USENET Newsgroup. Groups like rec.arts.books are very active and you may find readers online who could give you more details about a book that interests you. Thanks again for shopping at Amazon.com! Best regards, Lucille C. Amazon.com Earth's Biggest Selection httn://www.amazon.com/

Compatibility Selection System sm "How to Buy a Computer" PC Buyer's Preference Questionnaire

(Entered by office) Date of Questionnaire Code Number (DD - M M M - Y Y Y Y I BI Your name Your address Your Zip/PostCode Your telephone number Your fax number I I @I I I I I I I I I Your email address On boxes below, enter digit 1 to indicate first preference. Enter digits 2 to 9 to indicate lower preferences (duplicate entries permitted).

See "Chassis Type - Example" below.

CHASSIS TYPE - EXAMPLE DesktopIMini-Tower Mid-Tower Full-Tower Convertible 8 1 2 4 I 8 APPENDIX 5

Enter now your preferences...

CHASSIS TYPE Desktop Mini-Tower Mid-Tower Full-Tower Convertible

PROCESSOR TYPE Celeron Pentium II Pentium III AMD Cvrix

PROCESSOR SPEED (in MegaBerts = MHz) 300 to 350 to 400 to 450 to 500 to 550 to 600 to 650 & 349MHz 399MHz 449MHz 499MHz 549MHz 599MHz 649MHz over

MEMORY (RAM, in MegaBytes = MB) 32MB 64MB 96 MB 128MB 256MB 384 & over

HARD DRIVE (in GigaBytes = GB) 4 to 6 to 8 to 10 to 12 to 14 to 16 to 18 to 26 & 5.9 7.9 9.9 11.9 13.9 15.9 17.9 25.9 over

ROM DRIVE (Speed-X) CD CD CD CD48X CD- CD- DVD DV D DVD DVD32X 32X 36X 40X & over Writer ReWriter 4X 5X 6X & over

MONITOR (screen size in inches) 14 in 15 in 17 in 19 in 21 & over

GRAPHICS (video) CARD, in MB 4 MB 8 MB 16 MB 32 & over APPENDIX 5

OPERATING SYSTEM Windows95 Windows98 Windows2000

APPLICATION SOFTWARE Lotus MS MS MS MS MS MS MS Corel Smart Word Word Works Office Office Office O _"ice Office None Suite 97 2000 Suite 97 97-SBE 2000 20,S-SBE 2000

APPROXIMATE PRICE in US$ (VAT Tax included) $500 $1000 $1500 $2000 $2500 $3000 $3500 $4000 to to to to to to to & $999 $1499 $1999 $2499 $2999 $3499 $3999 over I I

On boxes below, enter 9 0 for the accessories you require and enter for the accessories you do not require. See "Accessories - Example" below.

ACCESSORIES - EXAMPLE Prin- Speak- Sound 56K Scan- Zip- Digital Joyter ers Card Modem ner Drive Camera stick Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No 9 0 9 0 9 0 9 0 0 9 0 9 0 9 0 9 Enter now vour requirements...

ACCESSORIES Prin- Speak- Sound 56K Scan- Zip- Digital Joyter ers Card Modem ner Drive Camera stick Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No

Compatibility Selection System sm COMPUTER DATING PREFERENCE QUESTIONNAIRE

(Entered b - Office Date of Questionnaire Code Number (DD - M M M - Y Y Y Y Your name Your address Your Zip/PostCode Your telephone number Your mobile tel. number Your fax number Your email address Your own sex Partner's sex Male/Female Male/Female APPENDIX 6

YOUR OWN CHARACTERISTICS On boxes below-right, enter digit 1 to indicate ArealLocation where you live. Leave all other options blank. See "AREA/Location - EXAMPLE" (belowleft) that indicates residence in the South West of the city.

AREA/Location - EXAMPLE AREA/Location where you live North North North North West North East West North East City City West Center East West Center East South South South South West South East W 1 9 --------------------------------------------------------------------------------------------------- 0#each characteristic below (AGE, HEIGHT, WEIGHT, etc.) enter ci -t 1 on the option that applies to you. Leave all other options blank. YOUR AGE 18-20 21-25 26-30 31-35 _j 36-40 41-45 4_6-50 51-55 56-60 YOUR HEIGHT in feet and inches 4'7" & less 4'8"-4/11" 5'0"-5'3" 5'4"-5'7" 5'8"-5'1 1" 6'0"-6'3" 6'4"-6'7" 6'8" & over 8 YOUR WEIGHT in pounds 80 & less 81-100 101-120 121-140 141-160 161-180 1+ 81-200 201 & over 8 YOUR RELIGION Jewish Protestant Catholic Mormon J. Witness Buddhist Moslem Agnostic 8 YOUR RACE YOUR LOOKS White Latin Oriental Asian Black Not Slightly Very Attractive Attractive Attractive Attractive 5 4 APPENDIX 6

YOUR MARITAL STATUS YOUR SMOKING HABIT Single Married Se arat. Divorced Widow. Never Rarely Sociall Often 5 4 YOUR EDUCATION YOUR DRINKING HABIT High-School College University Post-Grad. Never @arel Socially Often i 4 4 --------------------------------------------------------------------------------------------------- On boxes below, enter digit 1 to indicate the first preference that applies to you. Enter digits 2 to 9 to indicate your lower preferences (duplicate entries permitted). See "Taste in Food - EXAMPLE" below: Taste in Food - EXAMPLE German Italian Greek S anish French Chinese Indian Mexican English 6 1 7 2 3 54 9 5 Now, enter your own preferences: YOUR TASTE IN FOOD German Italian Greek S Danish French Chinese Indian Mexican English _ _ 9 YOUR TASTE IN MUSIC Rock Pop Jazz Country- Western Latin Semi-Classical ' Classical 7 YOUR PREFERED SPORTS _ Football Tennis Golf Swimming Squash Bowlin Skiing YOUR _HOBBIES Cycling Fishing [Dancing Cookin Garden in Travellin Readin YOUR OCCUPATION S Student Secretarial Clerical Sales i Craftsman Professional Man er Self-em toed 8 YOUR PERSONALITY Passive Sh Romantic Charming Magnetic Mature Dynamic Assertive I 8 ---- end of "Your Own Characteristics" ----- APPENDIX 6

YOUR IDEAL PARTNER'S CHARACTERISTICS On boxes below-right, enter digit I to indicate Area/Location where your ideal partner lives. Enter digits 2 to 9 to indicate lower preferences (duplicate entries permitted). See "AREA/Location - EXAMPLE" ( below-left).

AREA/Location - EXAMPLE AREA where partner lives: North North North North West North East West North East 4 9 9 city _ City West Center East West Center East 2 3 South South South South W .1 2 4 9 --------------------------------------------------------------------------------------------------- On each characteristic below, enter digit I for your first preference. Enter digits 2 to 9 to indicate lower preferences (duplicate entries permitted). See "Partner's Age - EXAMPLE" below: PARTNER'S AGE - EXAMPLE 18-20 21-25 26-30 31-35 36-40 41-45 46-50 51-55 56-60 7 3 2 1 2 6 9 i 9 9 Now, enter your preferences about your ideal partner: PARTNER'S AGE 18-20 21-25 26-30 31-35 360 4I-45 46-50 51-5556-60 9 PARTNER'S HEIGHT in feet and inches 47" & less 5'8"-5'11" 6'0"-6'3" 6'4"-6'7" 6'8" & over 8 PARTNER'S WEIGHT in ounds 80 & less 81-100 101-120 121-140 141-160 161-l80 181-200 20l & over I 8 APPENDIX 6

PARTNER'S RELIGION Jewish Protestant Catholic Mormon J. Witness Buddhist Moslem Agnostic 8 PARTNER'S RACE PARTNER'S LOOKS White Latin Oriental Asian Black Not Slightly Very i Attractive Attractive Attractive Attractive 5 4 PARTNER'S MARITAL STATUS PARTNER'S SMOKING Single Married Se arat. Divorced Widow. Never Rarely Socially Often 5 _4 PARTNER'S EDUCATION PARTNER'S DRINKING High-Sc hool Collea University Post-Grad. Never Rarely Socially Often 4 4 PARTNER'S TASTE IN FOOD German Italian Greek Spanish French Chinese Indian Mexican English 9 PARTNER'S TASTE IN MUSIC Rock Pop Jazz Country-Western Latin , Semi-Classical Classical 7 PARTNER'S PREFERED SPORTS Football Tennis Golf Swimming Squash Bowling Skiing 7 PARTNER'S HOBBIES _ Cycling Fishing Dancing Cooking Gardening Travelling Readin 7 PARTNER'S OCCUPATION S Student Secretarial Clerical Sales Craftsman Professional Manager Self-em to ed 8 PARTNER'S PERSONALITY Passive Sh Romantic Charming Magnetic Mature Dynamic Assertive 8 * * * end of Questionnaire APPENDIX 6

COMPATIBILITY SELECTION SYSTEM TABLE OF CONTENTS

Page (s) Applicant (particulars of) TECHNICAL FIELD 1/60 BACKGROUND OF THE INVENTION 1/60 Current chortcomings 2/60 Analogy 2/60 Illustrating the problems 3/60 Medical Diagnosis 4/60 Searching for Patents 4/60 Searching for books 5/6u Searching for Social Partners (dates) 6/60 Tender Buying 7/60 Inadequate Software 7/60 Internet Profits 8/60 BRIEF SUMMARY OF THE INVENTION 9/60 Advantages 10/60 Simulations 11/60 Solutions to problems 11/60 Medical Diagnosis 11/60 Searching for Patents 12/60 Searching for Books 14/60 Searching for Social Partners (dates) 15/60 Tender Buying 16/60

BRIEF DESCRIPTION OF THE DRAWINGS (FIGS. 1-55) 17/60 DETAILED DESCRIPTION OF THE INVENTION 21/60 Terminology 21/60 Object 21/60 Subject 21/60 Compatibility Selection 21/60 One-Way Compatibility Selections 21/60 Two-Way (Reciprocal) Compatibility Selections 22/60 Characteristics 22/60 Options 23/60 Intensity 23/60 Intensity Permutations 24/60 Primary Vs Secondary Options 24/60 Sumproducts 25/60 Compatibility Factor (CF) 26/60 Selection Engines 27/60 Website abbreviations 28/60 Example 1 - Selection of Personal Computers 28/60 Maximum number of selections 31/60 General Intensity Questionnaires for PC's 31/60 Completed Buyer's Intensity Questionnaire 31/60 Completed Seller's Intensity Questionnaire 32/60 Alternative "PC Buyer's Preference Questionnaire" 32/60 Controlled selection Runs A, B and C 32/60 Controlled selection Run A 33/60 Controlled selection Run B 34/60 Controlled selection Run C 35/60 Compatibility Comparison 37/60 The Perfect Match with CF=100% 38/60 The Perfect Match with CF=34.770 39/60 Reverse Selection for Seller S-491 39/60 Influencing Selection Outcomes Using Scalars 40/60

c,xample 2 - Selection of Social Partners (dates) 41/60 Maximum number of selections 44/60 General 'Computer Dating - Intensity Questionnaire' 44/60 Completed Male's Intensity Questionnaire 45/60 Completed Female's Intensity Questionnaire 45/60 Alternative "Computer Dating Preference Quest." 46/60 Controlled selection Runs D, E and F 46/60 Controlled selection Run D 47/60 Controlled selection Run E 48/60 Controlled selection Run F 49/60 Compatibility Comparison 51/60 Reverse Selections 51/60 Graphical Representation: The Compatibility Technique 52/60 Rare Selection Results 53/60 Unusual Score Result 53/60 Unusual Compatibility Factor Result 53/60 The TITAN Computer Program (and "The Master String") 54/60 The TITAN Source Code 56/60 TITAN1 Module 57/60 TITAN2 Module 57/60 CLAIMS 59/60 ABSTRACT 60/60 DRAWINGS (FIGS. 1 to 55) Sheets 1-122 APPENDICES: Appendix 1 email from J. Guerrero to Amazon Appendix 2 email from Amazon to J. Guerrero Appendix 3 sample of "3w.bestmatch.com" questionnaire Appendix 4 sample of "3w.matchmakerintl.com" questionnaire Appendix 5 blank "PC Buyer's Preference Questionnaire" Appendix 6 blank "Computer Dating Preference Questionnaire" Appendix 7 TITAN1 Fortran-77 source code Appendix 8 TITAN2 Fortran-77 source code

Claims

CLAIMS What I claim is: 1. A method for using a digital computer for the compatibility selection of Objects, such as goods or services, by comparing the Objects' characteristics (description) with a Subject's characteristics as described by the Subject, such as a consumer, comprising the steps of: (A) completing Objects' and Subject's intensity questionnaires for collecting numerical values, as expressions of human intensity (preferences), assigned to said characteristics of Claim 1; (B) a digital computer responsive to said numerical values of step (A) for generating intensity databases by incorporating the information collected on the said intensity questionnaires of step (A) ; C) a digital computer responsive to said intensity databases of step (B) for the one-way or two-way (reciprocal) compatibility selection of said Objects of Claim 1, by weighted sumproducts of the numerical values of step (A), with said Subject of Claim 1; (D) a digital computer responsive to said intensity databases of step (B) for the one-way or two-way (reciprocal) measure of the degree of compatibility in percent between said selected Objects of Claim 1 and said Subject of Claim 1; and (E) display means for exhibiting the results of said compatibility selection of step C) and said degree of compatibility in percent of step (D).

<Desc/Clms Page number 76>

Amendments to the claims have been filed as follows CLAIMS What I claim is: 1. A method for using a digital computer for the compatibility selection of Objects by comparing the Objects' characteristics (description) with a Subject's characteristics as described by the Subject comprising the steps of: (A) completing Objects' and Subject's in.%Gnsity questionnaires for collecting numerical values, as expressions of human intensity (preferences), assigned to said characteristics; (B) using a digital computer responsive to said numerical values of step (A) for generating intensity databases by incorporating the information collected on the said intensity questionnaires of step (A); C) using a digital computer responsive to said intensity databases of step (B) for the one-way or two-way (reciprocal) compatibility selection of said Objects obtained by forming, for each Object and each characteristic compared, the product of the Object's respective intensity value and the Subject's respective intensity value and summing the formed products; (D) using a digital computer responsive to said intensity databases of step (B) for the one-way or two-way (reciprocal) measure of the degree of compatibility in percent between said selected Objects and said Subject; and (E) using display means for exhibiting the results of said compatibility selection of step C) and said degree of compatibility in percent of step (D).
2. A method as in claim 1 where the Objects are goods or services and the Subject is a consumer.
3. A method as in claim 1 where the intensity values are weighted.