Cover Image

Not for Sale



View/Hide Left Panel

Descriptive Documentation

CHARLES G.SMITH

One of the advantages of having the proper materials and the ability to use them to make meaningful marks is that a record may be kept of what things are and where they are stored. Where there are many things to be stored, statement in the record of what they are requires distinction between them.

Where the disclosures of documents are concerned, “what things are” becomes a matter of their differing intellectual content. It is the discriminating assessment of this content for the purpose of document identification that poses a problem for which many solutions have been proposed.

An obvious approach to distinguishing between documents is to inquire in what respects some of the documents have disclosures that others do not. From the careful consideration of the thus found characteristics of the disclosures, a list, or a number of lists, of features, none of which applies to all the documents, may be organized upon some stated or readily apparent principle. By keying the list or lists to the location of documents it becomes possible to find documents having a desired disclosure by consulting the list or lists.

Library subject indexes and Patent Office classification schedules are lists of such type. Although they are designed to supply answers to different types of questions, each has the virtue that, if employed as designed, there is a group of documents isolated in which the answer may be found if it exists.

Word or subject indexes as used for libraries have the additional virtue that new subjects or topics relating disclosures to document location may be inserted easily. This is an ability not readily available in the substantially fixed schedules of classification. Patent Office classification schedules, on the other hand, have a certainty of locating documents which concern a specific relation of concepts that is not usually available in an alphabetical index of the library type. It is desirable in mechanized searching for the Patent Office that there be both ease of insertion of new combinations of concepts and certainty of finding of old combinations, if present, in the record searched.

While it is desirable that the valuable abilities of these systems be preserved,

CHARLES G.SMITH Office of Research and Development, United States Patent Office, Department of Commerce, Washington, D.C.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 1097
--> Descriptive Documentation CHARLES G.SMITH One of the advantages of having the proper materials and the ability to use them to make meaningful marks is that a record may be kept of what things are and where they are stored. Where there are many things to be stored, statement in the record of what they are requires distinction between them. Where the disclosures of documents are concerned, “what things are” becomes a matter of their differing intellectual content. It is the discriminating assessment of this content for the purpose of document identification that poses a problem for which many solutions have been proposed. An obvious approach to distinguishing between documents is to inquire in what respects some of the documents have disclosures that others do not. From the careful consideration of the thus found characteristics of the disclosures, a list, or a number of lists, of features, none of which applies to all the documents, may be organized upon some stated or readily apparent principle. By keying the list or lists to the location of documents it becomes possible to find documents having a desired disclosure by consulting the list or lists. Library subject indexes and Patent Office classification schedules are lists of such type. Although they are designed to supply answers to different types of questions, each has the virtue that, if employed as designed, there is a group of documents isolated in which the answer may be found if it exists. Word or subject indexes as used for libraries have the additional virtue that new subjects or topics relating disclosures to document location may be inserted easily. This is an ability not readily available in the substantially fixed schedules of classification. Patent Office classification schedules, on the other hand, have a certainty of locating documents which concern a specific relation of concepts that is not usually available in an alphabetical index of the library type. It is desirable in mechanized searching for the Patent Office that there be both ease of insertion of new combinations of concepts and certainty of finding of old combinations, if present, in the record searched. While it is desirable that the valuable abilities of these systems be preserved, CHARLES G.SMITH Office of Research and Development, United States Patent Office, Department of Commerce, Washington, D.C.

OCR for page 1097
--> it is also desirable that their disadvantages be avoided in applying them to mechanical search for Patent Office purposes. The use of word or topic indexing, although suitable for small collections of disclosures centered around some theme, does not appear to have the clarity of approach to specific relations of elements and the sure segregation of combinations of concepts, regardless of how they may have chanced to be expressed in the disclosures, that is essential in Patent Office searches involving many millions of documents from widely diverse technical fields. For mechanical searching the use of classified lists of possible aspects of subject matter has been proposed. Document identification then becomes a matter of applying to each document a selection from among the multiple aspects available in lists containing graduated groups of possibilities. A problem in doing this is the relation of the lists to one another. Where the variety of arts involved covers the whole of technology and questions requiring specific answers may concern any significant aspect thereof, the number of such lists needed may become unwieldy. Such lists tend also to have the undesirable rigidity now found in Patent Office Classification schedules which renders the introduction of newly discovered aspects of subject matter difficult. The proposed system is in contrast to such systems as may employ predetermined coding lists expressing various degrees of the probable aspects of documents. Rather it uses, in forming the record to be searched, a manner of organizing concepts such that the more complex concepts may be included in the record as relationships of elementary concepts. The proposed system, for example, does not apply functional names to identify devices or methods but spells out in the record the features, such as the change from one condition to another effected on the material treated, which in their interrelation necessarily characterize them. Any desired aspect can then be read from the record whether it can be named or not. The system thus provides a manner of describing disclosures rather than of classifying them by their aspects. It is thought that there can be found for each art a group of basic concepts by which persons skilled in the art describe documentary disclosures to each other. Such basic concepts are, in fact, now employed generally in the Patent Office in defining the meaning of classification schedules. From this viewpoint the problem of documentation becomes a matter of formulating a manner of relating concepts to one another, such that the basic concepts can be fitted together to form more complex concepts and they, in turn, fitted together until whole documents are comprehended. By suitably associating basic concepts, any significant aspect of any disclosure may be recorded and distinctions between documents would rest, as it does now, in the

OCR for page 1097
--> Patent Office as between classes and between subclasses, upon the relation of the basic concepts applied to one another. Technical documents are often alike in some respects but differ in others. They, furthermore, differ in the manner in which they include the same idea. In order that the disclosures may be brought to a common basis for comparison, a system of analysis is proposed. The analysis of the document serves as an intermediate stage between the disclosure and the coded record in that it renders the disclosure into the concepts and relations to be used in the coding. The coding to form the record and the mechanized search thereof must be in the light of the abilities available in some machine. As herein described the system is designed for use with the Interrelated Logic Accumulating Scanner (ILAS) of the U.S. Patent Office (1). Use with other machines is a matter of adaptation to their abilities. It is thought that by the use of this system of describing disclosures the introduction of a new topic or aspect would be readily accomplished since it would be merely another organization of available basic concepts. Furthermore, since it relies upon the relations of the same ultimate basic concepts as now employed for the purpose, the ability to specify detail as needed is also available. A descriptive system Since the disclosures of technical documents concern physical things, one way of identifying them in a record is to apply to them various names or terms indicating what things are present in each document. The mere recitation of the terms representing the things present is not enough, however, to discriminate sufficiently between documents where large numbers of documents are concerned. An inventory of the parts present would be of little value in distinguishing disclosures for the type of search required in the Patent Office. It is the relationships between things that are material in the search for specific devices. Many devices may have substantially the same parts, but they bear little other resemblance to one another. A system to be appropriate for the novelty type of technical search should be able to relate terms to each other so that devices of the same components may be distinguished. Relationships between terms are not themselves terms and cannot be satisfactorily treated as though they were. A relationship between things is a sort of abstract form into which various terms may be placed. It is necessary, hence, that relationships not only be present in the record but also that the record be clear as to which terms are involved in each relationship. Where in all the documents for which a record is to be made there is the

OCR for page 1097
--> same predetermined, or inherent relationship between things, it may be feasible to distinguish the documents by naming the things present in the light of such omnipresent relationship. The naming may be made more specific by defining species for each of the things partaking in the basic relationship or combination. Within the disclosures of each patent, however, there are many combinations or relationships that must be in the record. Where all the documents do not present a fixed relationship between the things disclosed, the system must have some way of including relationships as well as the terms and it must have some way of referring the terms to the particular relationships in which they are involved. The inclusion of a relationship should be made in such a way that various terms may be supplied to it in a uniform manner. One way (2) of including a relationship in the record and linking terms to it is to apply the relation to each term involved and then connect the terms to one another. Doing this may result in treating relationship as though it were a description of a thing rather than a relation between several things. Furthermore, each thing is involved in many relations to others. It may, in fact, have opposite roles in the same relationship as applied to different other things, and ambiguity may result. Relationship abstractly and basic coding patterns Another way of including various relationships in the record and specifying the terms to which each applies is to express the relationship apart from the terms and then join terms and relationship to form a combination. This manner of expressing relations is followed in the proposed system. An advantage of this expedient of the present system is that the ability to establish a chain or sequence of relationship is secured. Such a sequence may be expressed in various ways; for example, a R b b R c c R d where a, b, c, and d are conceptually distinct elements and R is a relation that exists between them. The same set may also be written or diagrammed R: a→b→c→d

OCR for page 1097
--> From this basis it is desirable for search purposes that the record be readable a R d even though there may be unknown intervening relations as to other elements. Such a result is available through use of the present system. The coding for retrieval by ILAS which makes this result possible is developed as follows: where M is a pulse available for the row in which M occurs. By the use of a different column in the interfix area for each set the record becomes: The relation, as applied to any two terms X and Y where they are wanted only if directly related, may be recovered by this hookup for ILAS: Note that the M pulse applied to the relation is used as a relay opening pulse. The sets may be related to one another by extending the relation interfixing of preceding sets to the columns directed to the other sets:

OCR for page 1097
--> This is the record for a chain relation where a precedes b which precedes c which precedes d. The a R d relation in the record stated is indirect. In order to recover it if wanted whether direct or indirect the following hook up is used: The record is read by ILAS from the top down as shown for the chain relation. Thus with the hookup last shown the finding of “a” will close a relay by pulsing the “pickup.” When the immediately subsequent R appears pulsed by M, a hit is scored in the interfix field and the relay is “dropped out” or opened. This hit will be in all three columns of interfixing pertaining to the topmost R, as shown in the record. Subsequently when the “d” appears and is found interfixed in one of these three columns, a final hit is scored. A frequent situation which arises in various aspects may be diagrammed: The same hookup or machine instruction as given previously with reference to the chain of relations is employed for retrieval for this branching chain. The circuitous situation is also of major importance:

OCR for page 1097
--> Here, too, the same machine instructions or hookups used in the retrieval of the chain are applied. Concepts The diagrammatic analysis and coding patterns developed thus for expressing relationships are useful in recording a wide variety of situations as found in technical disclosures. They are being applied in this project to methods and machines. It is, of course, necessary that there be intellectual concepts which lend embodiment to these abstract forms. Another aspect of the present system is the use of a few broad concepts that are widely applicable and commonly present in many arts. The system reaches for the ultimate concepts which are commonly employed in language and easily understood. These concepts are those necessary to the definition of more specific concepts. A meaning involved in a more specific concept is secured by the use of the power of interrelating broad concepts to each other. The concepts to be applied to documents may be definable. If they are definable it is in terms of more elementary concepts which must finally have meaning only as based upon actual experience. Thus there is a basic stratum of concepts which need no definition. It is the use of such elementary concepts that is contemplated in the present system. As but a limited number are used and they differ widely, the matter of the choice of which concept to apply is simplified. An advantage of a documentation system capable of expressing combinational relations is that the elemental concepts employed may be few in number and of a very broad and easily understood nature. This advantage arises in that the distinction between documents would not depend upon the use of a voluminous and highly discriminated terminology, such as is often employed in categorical classification systems, but would rest upon the manner in which simple concepts were organized. The ability to combine concepts implies the ability to express complex ideas by suitable composites of simple ones. Many words as employed in natural languages represent combinations of more elementary concepts. Such words are a great convenience in spoken or written communication but their use in coding for documentation is undesirable since it places reliance upon such of these words, and their then meaning, as chance to exist at the time of the creation of the codes. Tomorrow may see other words employed for the same composite meanings. Indeed, today, in various arts, different names or words are employed for substantially the

OCR for page 1097
--> same complexes of concepts. Furthermore, there are some combinations of elementary concepts for which no name presently exists perhaps because the circumstances, such as invention, necessary to foster creation of such words have not yet occurred. The use of summarizing words is unnecessary if the combinations of elementary concepts for which they stand can be expressed directly. It is fundamental in the present system to go behind composite words to the basic organization of elemental concepts which they represent and to set forth the combination essential to the definition of such words. Where the power of organization is available, complex or composite concepts can be reduced to their elements since the elements may then be organized to express the composite concept. It may be feasible, as to particular arts, to analyze the facts involved and concepts employed, against the background of a system of possible logical patterns, to arrive at a small group of basic concepts in terms of which all the combinations present in the documents may be expressed. So far as the combinations in them may differ, the documents can be distinctively identifiable, even though the elements and relations per se in them may be the same, since an indicated relation may apply to different elements in each document. Perhaps a fundamental array of concepts can be found and a common manner of organization employed to the end that all technical literature may be viewed in a common light. It is to be noted that the present Patent Office Classification Schedules are composed of specific terms which are relations of more elementary concepts. These class schedules are accompanied by definitions setting out the defining combinations of elementary terms. In view of the specificity thus obtained the present use of organizations of elementary concepts is thought well founded. What concepts in the total that there will be finally cannot be forecast. The list of concepts must grow by the study of particular arts and the carry over from one art to another. The few concepts necessary to any particular art will be easily grasped and understood, as they are now relative to classification schedules, by those who search the art. Method disclosures It is highly desirable where a great number of documents are concerned, as in the present case where there are millions of patents involved, that there be some common foundation or thread running through all of them to which resort may be had to coordinate for various disclosures. If some common basis

OCR for page 1097
--> can be established from which all documents may be approached, then it is feasible to relate in the record each disclosure to all the others. The linking factor among patents is the method or operation disclosed. All inventions to be patentable must have utility. Utility involves necessarily the use of the invention in some method. Thus all patents concern method. The primary reason for the existence of machines and devices is the function they perform. Apparatus is created because it has utility in the performance of some method. Apparatus is, in fact, usually denoted by either (1) the activity it causes in some material, such as “agitating,” or (2) the state it produces, such as “vaporizing.” Furthermore, methods are themselves patentable subject matter. Every class, or nearly every class, in the Patent Office, has, at present, patents directed to methods and usually these method patents are provided for in the class schedule. It therefore becomes essential to have a concept of method which will furnish a foundation for making a documentary record suitable for patent search. In the proposed system a method is considered to involve a sequence of states and the change from one state to another. The record may thus be formed by expressing the states of a material or of a system of things at successive times and indicating that there is a change in a specified direction from one state to another. A state may be defined as any condition, status, or activity which can be observed or experimentally determined. It is a sensible aspect of things. It refers to those evidences there are of what a thing is which may be determined by observation and testing. To express it, only ordinary concepts, as used in defining technical viewpoints, are necessary. These concepts are the basic concepts by which complex ideas may be defined. Method is considered as a sequence of states. Entity: S1 → S2 → S3 → S4 R: Sx → Sy=change of state from Sx to Sy Sx=state of an entity Method analysis involves Successive states that are recognizable if they occur: a material or object treated or of a system of entities The concepts needed to define further a specific process are those also needed to set forth the state or condition of the thing. These will depend on the specific process or art involved. As it is the purpose of this paper to outline a

OCR for page 1097
--> documentation viewpoint and procedure rather than deal comprehensively with the arts, no attempt is now made to state all the concepts of state that may be concerned here. Machine disclosures Machines are viewed, functionally, as means for producing a change of state or an effect and for applying energy for that purpose. In producing a change, the operation of the machine is related to the method. It may be said that the machine takes in something in one state and puts it out in another state and that what it takes in includes energy in some form. Machines have many aspects from which they may be considered. Among them are: Machines of different function may be used in conjunction. There may be a series of devices employed successively to treat the same entity or material with each device making a different change in it. The several machines may or may not take their operational energy from the same source. Machines functioning for the same purpose may employ different structures for doing so. It is desirable that the record be of such a nature that these structural differences may be included in proper relation to the operational aspect with which they are concerned. Even though they function for the same purpose and employ the same structure for doing so, machines may differ from one another by the presence of devices which modify the operation of the machine so that it may better fulfill its purpose. Among such added devices are controls which modify the operation in response to a detected condition or change of conditions. Such devices use energy and such energy may come from the same source as the operating energy, or it may come from some other source. Machines functioning for different purposes may yet have components which are alike or which serve the same subsidiary purpose in both. These components should be identified in the record in similar manner wherever found. Machines must be regarded as both static devices and as operational devices. They may be looked upon as functioning for some purpose as well as being structures the parts of which support and constrain one another. As the latter they must be considered along with other such structures that are not machines in designing a search system. As it is typical of machines to employ energy, it is thought fitting that as an example of its application to accompany this description of the system an activating means should be employed. That chosen is a hydraulic servo motor

OCR for page 1097
--> system. The choice was guided, in part, by a present need for mechanizing search in that field if feasible. It was also guided by the view that a system suitable for this art would be applicable specifically in other fields such as brakes, presses, pumps, and fluid motors. Operational Analysis Energy is generally considered to have various forms and to be convertible from one form to another. These forms may be regarded as the states or condition which energy may have or assume and the conversion of energy from one form to another as a change of its state. A series of changes of the form of energy may be diagrammed analytically as was indicated for a method disclosure above: Energy: S1 → S2 → S3 → S4 Energy is not, however, a directly recognizable thing but a concept which is evidenced in the state or condition of some entity. Some of the forms in which energy may be evidenced are the flow of electric current, the flow of a fluid, and the motion of a body in rotation or rectilinearly. In each of these “states” of energy there is some activity of an entity. Energy in action has the nature that when applied to a device it produces some necessary consequence. When energy is put into a translating device, it produces as a result the output of the device. In the present system this relation is regarded as “with the result that” which is abbreviated as WRT. Let us suppose we have three devices, D1, D2, D3, each of which receives energy in one form and converts it to another and, further, that these devices act serially so that the output of one device is the input of the next following device. The devices may be observed to have structural portions which act and co-act as indicated: D1 has two portions: A static portion having the connections to which the electric current is supplied. A rotary portion having a shaft which rotates as electricity is supplied. D2 also has two portions: A static portion with a chamber having connections to one of which liquid is supplied and from another of which liquid flows as the shaft below is rotated. A rotary portion having a shaft which when rotated has the result that the liquid flows.

OCR for page 1097
--> D3 has two portions: A static portion having two chambers separated by a movable wall, a connection to one chamber receiving liquid, a connection to the other chamber discharging liquid. A piston forming the movable wall dividing the chamber of the static portion into two separate chambers and with a projecting rod that moves rectilinearly out of the static portion as the liquid enters one of the chambers. This disclosure may be analyzed by diagram: In this diagram the vertical arrows point to an activity resulting from another activity. In addition to these resulting activities other features present include: The increase in pressure of the liquid during flow through D2. The transfer or travel of the liquid from place to place during operation of the device. These have also been symbolized in the analysis. Coding Operation The basic coding pattern to be applied to this analysis may be diagrammed: In accordance with the basic coding the coded record for the devices with regard to energy is:

OCR for page 1097
--> Note the placing of the entity involved in a row preceding that in which its activity appears. The only change in hookup required by this for search will be that the entity must trigger a relay to pass the immediately following state of activity of the entity and the relay then be reopened. The entities considered to be present may be listed as: These entities may now all be included in the coding:

OCR for page 1097
--> The three devices were indicated to be “transducers” taking energy in and putting energy out in either the same or a different form. These devices may be bracketed to indicate the input and output states of energy for each. This may be done by applying a like signal for each terminal state. Which is input and which is output is determined by the interfixing. The same signal must be used because the output of one is the input of the one next following. The record now becomes, using an 8 signal for this purpose: Note in this arrangement that the input state is at the top of the recitation for each device and the output state at the conclusion. Note, too, that detail of the D2 stator, a chamber, is on the same line as the stator which it features. An additional device may be added in the form of a supply chamber, D4, or “reservoir” for the liquid which goes to D2 and then to D3. With such addition the flow of liquid may be included. In some devices using this arrangement (servo motors) the liquid2 coming from chamber2 also returns to the “reservoir.”

OCR for page 1097
--> The liquid flow may be diagrammed: Liquid1 flowing: D4 → D2 → D3 Liquid2 flowing: D3 → D4 The flow of liquid is coded by placing “to” in the row following the source and interfixing both rows with “flowing” and the destination. Where the flow is to several chambers successively a coding chain appears. The flow of liquid may now be included in the coding of the example: A third aspect of the operation is the change of state of the materials concerned. In the instant device this is represented by the increase in pressure of the liquid by D2. Such change of state is coded in a third interfix area.

OCR for page 1097
--> For the pressure change the coding is: The linking of the interfix areas is by the use of the same row for aspects of the same concept. Controls Analysis Devices of a general basic operational design may differ from one another by the controls which are applied to them. Servo motors differ in this respect as to the control of the flow of liquid. A control is related to the operation of a machine in that some change in the relative position of the parts of the control has the result that there is a change in the operation of the machine in some way. The change of position of a valve, for example, results in a change in the amount of fluid flowing in a system. In one aspect (“stop-go”) a control may have only two states in one of which the operation controlled is permitted and in the other of which the operation is prohibited. In another aspect of a control device the operation controlled is always permitted but not always at the same rate or in the same degree; the control has the two states of “more” and “less” and change may be in either direction. Several controls each of “stop-go” aspect are often employed in conjunction by linking their operators mechanically together. The usual objective of this is the ability to involve different actions either alternatively or alternately. This is done by requiring one control to be “go” when the other is “stop.” Controls are always operated by the use of energy. The source from which the energy is drawn may be muscular, as in manual control, or it may be of another form. A gravity bias may operate a control; a spring may, too. These energy sources represent energy other than that supplied operationally to the system. An important source of control operating energy is from the operational system. The system itself may be tapped somewhere so that there is a diversion of a portion of the energy supplied from a source which supplies the operating energy.

OCR for page 1097
--> Control operation may be made contingent upon some aspect of the operation of the device controlled. Control of the device by its own internal conditions involves some means which varies with or detects an operating condition of the machine. It also involves a relation of the change which occurs (and which is detected by a change in a detector) to the change made in the control device. Hence with control by machine condition there is a sequence of change relations. For the analysis of control a simple symbolism may be applied to the basic analytic diagram. In a hydraulic system, for example, a valve may be shown

OCR for page 1097
--> simply as with an indicator for a hand wheel if manually actuated or suitable pressure chambers and/or springs if so operated. Other control devices may be given similar symbolism. Each of these symbols represents a rather standard coding group of entries based upon the relation of a control to the operation of the device. Coding of Controls In the coding of the flow controls, it is necessary to indicate the relation of the change in the control to the change in the flow activity. It is also necessary to include the energy source which acts with the result that there is a change in the control. The coding for a manually operated stop valve in operative relation to a flow system is given herewith. Other inputs of additional energy to the system for control purposes, such as bias springs or gravity, would be included as is the manual energy in the example. Where the control is from an operational change in the flow system a detector is involved which changes as a result of the change in operation of the system. The change in the detector in turn results in a change in a flow controller. The following coding is for the change in the controller of flow in one line

OCR for page 1097
--> by the change in pressure in another line to the extent that an increase in pressure in the controlling line (liquid1) has the result that flow is permitted in the controlled line (liquid2). This situation arises in servo motors where the application of pressure to an input line to one chamber causes an outlet valve to open in the output line from another chamber. Searching the coded record Any of the chains of relations in the various interfix areas may be searched by substantially the same hookups as given previously relative to the basic coding patterns. There is additionally involved further dependence of one row upon a subsequent row in the same manner as one row was made dependent upon triggering by a prior row in the basic hookups. Involved, also, is the intersection of cumulative hits registered in plural interfix areas before a final hit is obtained. This requires merely the closing of separate relays by the several cumulative hits with the requirement in the

OCR for page 1097
--> hookup that they all be closed before the final pulse may pass. This is a technique familiar to those who use relay type machines. It is to be understood that these hookups might be included here but, if they were, they would be with reference to the experimental ILAS machine and of little significance since it has insufficient relay capacity to handle the hookups needed. It is hoped that a production machine will be shortly available. Conclusion It will be noted that the concepts employed in the analysis and coding are few and of a very broad nature, yet they are so related by the coding that the ability to specify combinations of factors to any desired degree is available REFERENCES 1. ANDREWS, DON D. “Interrelated Logic Accumulating Scanner (ILAS), Patent Office Research and Development Report No. 6, Department of Commerce, Washington 25, D.C., 1957. 2. Ibid., pages 8, 9.