Cover Image

Not for Sale



View/Hide Left Panel

The Possibilities of Far-Reachmg Mechanization of Novelty Search of the Patent Literature

G.J.KOELEWIJN

The Netherlands Patent Office has undertaken research into the possibilities of far-reaching mechanization of novelty search in the patent literature, starting with the subject, carburettors. When the outline of a paper relating to the carburettor study was drafted, it was expected that by the time the paper itself was to be submitted a preliminary test would have been performed with the available material. Unfortunately this has not been possible. Barring unforeseen circumstances we are quite confident that within 3 to 4 months, i.e., well in advance of the conference, the system will have been put to a test. In view of the special nature of the selection means that have been developed during this research, I am of the opinion that despite our failure to report the final results achieved, a paper describing our work could be important enough for the conference, and I have decided to submit a report on our progress to date.

Aim of the carburettor project

The most important aim of this study is to obtain a better insight into the possibilities and difficulties with regard to mechanical selection of patents for the purpose of the novelty search. The desirability of such studies for the U.S. Patent Office is admirably brought out in the Bush report (1). We were of opinion that a mere study of the literature on the subject would be quite inadequate for the purpose. It was considered necessary to get some experience of our own. In view of there being only a few members of the staff available to pursue this object, a certain choice had to be made. There were several reasons for choosing a subject in a mechanical engineering field:

  1. In this field the shortage of staff was least in evidence.

G.J.KOELEWIJN Member of the Board of the Netherlands Patent Office, in charge of the Classification Department and the Library, The Hague.



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 1071
--> The Possibilities of Far-Reachmg Mechanization of Novelty Search of the Patent Literature G.J.KOELEWIJN The Netherlands Patent Office has undertaken research into the possibilities of far-reaching mechanization of novelty search in the patent literature, starting with the subject, carburettors. When the outline of a paper relating to the carburettor study was drafted, it was expected that by the time the paper itself was to be submitted a preliminary test would have been performed with the available material. Unfortunately this has not been possible. Barring unforeseen circumstances we are quite confident that within 3 to 4 months, i.e., well in advance of the conference, the system will have been put to a test. In view of the special nature of the selection means that have been developed during this research, I am of the opinion that despite our failure to report the final results achieved, a paper describing our work could be important enough for the conference, and I have decided to submit a report on our progress to date. Aim of the carburettor project The most important aim of this study is to obtain a better insight into the possibilities and difficulties with regard to mechanical selection of patents for the purpose of the novelty search. The desirability of such studies for the U.S. Patent Office is admirably brought out in the Bush report (1). We were of opinion that a mere study of the literature on the subject would be quite inadequate for the purpose. It was considered necessary to get some experience of our own. In view of there being only a few members of the staff available to pursue this object, a certain choice had to be made. There were several reasons for choosing a subject in a mechanical engineering field: In this field the shortage of staff was least in evidence. G.J.KOELEWIJN Member of the Board of the Netherlands Patent Office, in charge of the Classification Department and the Library, The Hague.

OCR for page 1071
--> Mechanical topics are as a rule more intelligible to those who are inexperienced in the given field than chemical or electrotechnical topics. Experience is thus more easily gained. There already exists an extensive literature on mechanical selection in the chemical field, but there is very little relating to mechanical or electrotechnical topics. The problem of mechanical selection of electrotechnical circuit diagrams appeared too difficult to begin with. Carburettors were chosen for the following reasons: In regard to this field we are convinced that by the usual documentary methods it is not possible to achieve a fairly extensive subclassification that is really satisfactory. The aspect “shape” which is difficult to define for purposes of mechanical selection plays no important part here. Here as with many other subjects that are difficult to classify the emphasis lies in the constant combination, in different relation to each other, of a limited number of detailed characteristics that are not very special when considered separately. From the point of view of classification there appears to be a high degree of correspondence with electrotechnical circuit diagrams. Documentary systems in use at the Netherlands Patent Office THE SYSTEMATIC COLLECTION OF PATENTS At the Netherlands Patent Office the systematic collection of patents comprises 5 million patents (Belgian, British, French, German, Luxembourg, Netherlands, Swiss, and U.S.). Since 1946 other publications or abstracts thereof have also been incorporated in the collection. All these documents are arranged according to the headings and subheadings of the Netherlands classification system. In principle, each document is classified under the symbols of one heading or subheading, but may be allotted symbols of one or more other (secondary) headings (or subheadings). On the average, about 130 secondary headings are in use for every 100 Netherlands patents; at present about 40 secondary headings are in use for every 100 recent foreign patents, the extra copies called for being ordered specially or reproduced photographically. One great advantage of such a collection is that after determining the headings and subheadings of the system relevant to a novelty search, the examiner can at once consult the patents and other documents filed together in the appropriate

OCR for page 1071
--> filing cupboards. The classification system for such a collection in which as far as possible each document is filed under a single symbol, is, naturally, to a high degree adapted to this purpose. The same applies to the mode of classification of the patents—attention is focused primarily on the novel concept disclosed therein. Since the very purpose of the existence of a patent is to disclose and define a single inventive and/or novel concept, an experienced worker can as a rule quite rapidly form an impression regarding the essence of the invention, for which purpose usually only one or a few points of view are of real importance. In consequence classification of such patents in the Netherlands system does not take much time. On the average, well-qualified personnel, with experience of classification work, require only 7 minutes per patent for the complete operation of classification. Considering that the Patent Office acquires each year some 150,000 new patents that are incorporated in the classified collection, the classification of these demands the equivalent of the annual working time of 10 experts (the actual number engaged on this work, as part of their duties, being far greater). Many objections can be levelled against this system. On the other hand it has the great advantage that it requires relatively very little time of technical experts and that the patents themselves are arranged in systematic order. In all technical fields in which with this system a novelty search continues to afford reasonable results in a reasonable time, it is apparent that the introduction of any other system cannot be justified on economic grounds. Here one has to take into account the annual total of patent applications; what may be uneconomic for our patent office may well be feasible for a much larger organization or in the event that another system is brought into being on an international scale by a cooperative effort of a number of organizations. One of the most obvious objections to a system in which the patents are arranged in groups is that the steady growth of material (there is constant evidence that old patents can be important) makes constant subdivision of the groups necessary, so that in most cases the limit has already been reached at which the subject matter of a patent falls completely within the scope of a particular group of lowest hierarchy. Moreover, classification according to the (presumed?) inventive concept involves a limitation that has certain practical advantages, but it is nevertheless a serious limitation, for it results in much, sometimes very important, information contained in a patent being to all intents and purposes unavailable for purposes of novelty search. After all, what matters for a novelty search is not particularly the scope of the claims in the published patents, but everything that appears therein. One will of course always have to make a choice in regard to what information is to be traceable

OCR for page 1071
--> via the documentation system, since the incorporation in a traceable fashion of all kinds of information that is already well known from other sources is of no value in itself, and can easily lead to overdocumentation. SPECIAL SYSTEMS In any event, there has for a long time existed a need for multiple classification. In the case of a subclassification within a small self-contained group of patents (e.g., a group that has become inconveniently large) this can be achieved without sacrificing the possibility of arranging the patents themselves; the latter are analysed in the light of the subdivisions envisaged and are marked with one or more symbols accordingly. Patents similarly annotated are then grouped together and these various groups are then distinguishably arranged in a definite order in accordance with a chosen hierarchy for the symbols and their combinations. This little known possibility has turned out to be most practical in certain cases. If the multiple classification extends to a rather large number of aspects, separate auxiliary means are called for. Since about 1930 we have made use of what we call combined lists in the fields of automatic telephone systems and of electrical control mechanisms. These lists include a column for patent numbers, a column for the classification symbol under which a patent falls in the normal system, and a large number of columns in each of which the presence of one or more aspects can be indicated by appropriate symbols. On the same line as the patent number, definite symbols are thus recorded at definite places to indicate the aspects presented by the patent. With such a list it is possible to delegate to administrative personnel the task of retrieving patents corresponding to a given set of aspects; this task involves working through the lists to identify the relevant patents and removing the latter from the collection and delivering to the examiner the combination group that has thus been temporarily constituted. It seems to me that marginal selection card systems, Batten systems, and punched-card systems, in all of which each perforation has a unique significance, can offer little advantage in principle over such systems based on lists. On the other hand, such mechanical systems enable the search by the administrative personnel to be conducted more rapidly and with greater accuracy. In addition, marginal selection systems offer the possibility of recording summaries of the document on the cards themselves, while in the Batten system the choice of the aspects is easier to modify. Finally mention will be made of a card index that is in use for the azo dyestuffs. To start with, two cards are prepared for each substance, the index being duplicate; for each substance a card is prepared corresponding to the diazo compound and another card is prepared for the coupling component.

OCR for page 1071
--> The structural formulae are represented by a code. The code consists of two parts, one for the carbon skeleton and the other for the substituents. In both cases a relatively simple code is used, which does not express the structural formula in every detail. This is done deliberately, since we are here concerned not with allotting a unique code to each structural formula, but with the construction of a system which permits rapid retrieval of documents pertinent to structural formulae of a given type. The code for a structural formula is thus composed, as it were, of code units in such a way that the units remain recognizable. Description of a simple example will give some idea of the code system: is alloted the code C 11:10–4–610, in which the units have the following significance: C carbocyclic compounds C 1 ditto, with carbon skeleton composed of rings that are not connected by carbon chains C 11 benzene : precedes the code for the substituents 10 —OH (1=singly bound oxygen; 0=hydrogen) – separates the substituents 4 halogen 610 —SO2—OH (6=—SO2; 1=singly bound oxygen; 0=hydrogen) A definite order of precedence is laid down for the various components of the code, and the cards are arranged in this order in such a way that a given compound comes in a definite place and can be looked up directly. The cards are interspersed by guide cards designed according to a carefully chosen system. In the example above, the corresponding cards are located behind the fourth guide card governing benzene derivatives. Thus no. 1 refers to code C 11, no. 2 to 10, no. 3 to 4, and no. 4 to 61. The guide cards make it possible to look up any desired combination of the individual code units, either alone or in combination with other unspecified code units. To this end the sequence of guide cards is determined behind the last of which the desired combination would be situated, and one then proceeds to locate the guide cards concerned. For example, if one has to look up a phenolsulphonic acid, the code for which is C 11:10–61, one encounters the relevant cards behind the guide cards

OCR for page 1071
--> C 11, 10, and 61. Having discovered guide card C 11, and the existence of a guide card 10 for the first substituent, one looks to see if guide cards subsidiary to the latter are in position for a second substituent (the protruding tongues of the guide cards are situated at different positions along the top edge, and cards bearing a code for a first substituent are arranged with their tongues in line while cards designating a second substituent are chosen with tongues slightly displaced relative to those of cards denoting a first substituent and so on). One inspects successively the guide cards for a second substituent subordinate to 10 until one reaches guide card 61, behind which will be found cards relating to phenolsulphonic acid. Subordinate to this guide card may be found guide cards for a third substituent of lower hierarchy than 6 (e.g., 7, 8, or 9). If a third substituent is of higher hierarchy, e.g., 4 (halogen), the substance (halogenophenolsulphonic acid) is found further forward in the system, namely behind third substituent guide card 61, subordinate to second substituent guide card 4, subordinate in turn to first substituent guide card 10. If one is interested in a phenolsulphonic acid in general, irrespective of other substituents, a series of positions in the system will thus have to be inspected, but within a relatively short range. The system embraces 6,000 patents, for which some 30,000 cards have been made; the search instructions can be formulated very precisely, and the search can occupy anything from a few minutes to at most, an hour. A disadvantage of this system is that patents in which many substances are mentioned require a considerable expenditure of time for coding, and involve a large number of cards for their incorporation in the system. For the rest, it is a system that meets the most stringent requirements. A characteristic (structural formula) is coded in terms of units in such a way that a search can be instituted for any desired combination of units. The use of one card for each characteristic guarantees that the various units into which a characteristic is broken down are not simply all mentioned somewhere in the patent recorded on the card, but that they are mentioned in combination as that characteristic. Special features of the search for novelty in patent specifications In order to obtain a good understanding of the reason that led to the system described subsequently in this paper, it is necessary to consider for a moment more closely the special character of the search for novelty in patent specifications. This special character is a consequence of the typical features of patent specifications as well as the search for novelty.

OCR for page 1071
--> THE PATENT SPECIFICATION A patent specifies a right granted by a state to a person or corporate body, based upon the patent law of that state. It is therefore a document with a juridical character. Usually a patent law requires the applicant to describe and explain his invention in an application for a patent in such a way that his description enables a person skilled in the relevant art to reproduce the invention. This description is then incorporated in the specification of the patent when it is granted. In many cases the applicant considers this requirement to be contrary to his own interests and the descriptions will therefore often be at the borderline of what is allowable in this respect. The juridical character of the patent specification results in the use of unusual terms in describing the invention, so as to cover all foreseeable alternative technical possibilities, and also the contention that the invention is useful for many applications (that usually have not been tried); in short, in a very broad and technically not very detailed picture of the invention. All this is accentuated by virtue of the fact that, first, in the interest of prompt filing of the applications the inventions are described at an early stage of their development when often a good terminology in the art is not yet available, and, secondly, the specifications are drafted by patent experts, each of whom has to cover a wide technical field and consequently often has difficulties with applications in special technical fields in which he is not technically expert. On the other hand, since patent laws usually require applications for patent to be restricted to a single subject of invention, patent specifications have the character of monographs, which is advantageous from the point of view of classification. To summarize, a patent specification is thus mainly a monograph in which the accent lies upon something new that is described, in which a fairly sharp formulation of the exclusive rights claimed is contained, but of which the contents, from a technical point of view, are not always clear and in many cases not completely justified. THE SEARCH FOR NOVELTY When an examiner examines an application for its patentability, he finds in it a combination of characteristics which the applicant considers in the first place to be new and in the second place so different from what is already known that the normal person skilled in the relevant art would not automatically hit upon this combination. These are two of the most important requirements for patentability in most countries having an examination proced-

OCR for page 1071
--> ure. The examiner must verify that these requirements are fulfilled, and will for that purpose among other things conduct a search for novelty in patent specifications. If it emerges that the combination of characteristics sought has already been described in a single patent specification (that was published before the filing date of the application), then this constitutes a bar for granting another patent on this combination. In that case the examiner has no interest in finding any other specifications. If no patent specification is found containing the full combination of characteristics sought, the examiner has to collect sufficient material to obtain a clear picture of the state of the art, in order to be able to assess the level of invention of the application. He will be able to do this if he selects some less restricted combinations of characteristics that still give a good approximation to the invention, and then studies all patent specifications found for these combinations. It is here the desire to locate all specifications describing the selected combinations, just as with literature searches for other purposes. As the foremost feature of mechanical search for novelty I see the very great importance of the possibility of making a very sharp selection, combined with the possibility of widening the search in very small steps (4). Next it seems to me that also the nature of the combinations of characteristics searched for strongly bears the mark of the search for novelty, and there are no restrictions with regard to applicability in practice. This also influences the nature of a selection system. Requirements to be fulfilled in a very high degree by a mechanical selection system for search for novelty in patent specifications First, it must be possible to locate relevant patent specifications irrespective of unclear or unusual terminology in their text. This requirement can be fulfilled in a much higher degree by a system based on selection of notions than by a system based on selection of words (as, for example, in a Uniterm system). Secondly, because in formulating inventions notions are used that are as wide as possible, and because each generic notion searched for (in combination with other notions) is fully present in all its species, it is extremely important that when searching for a wide notion all its species are also found. Therefore the use of a code that expresses the essential hierarchy is very important (3, 5). In systems in which this is not possible (Uniterm system), either, when formulating search criteria, all possible species of the notions to be searched for must be taken into account or, when analysing and indexing specifications for

OCR for page 1071
--> each characteristic contained, all its generic characteristics must also be brought into the system. In both cases this is quite a problem. A mechanical system will only justify the work put into it and the inconvenience caused by the inevitable additional steps between the formulation of a search order and the study of the specifications found, if a very sharp selection on restricted subjects and a wide variation in search orders is possible (see, in this respect, “The search for novelty” above). The limits of the field of technology covered by the subject of the investigation “Carburettors” The subject of investigation is limited to what could be called the more or less ordinary carburettors for internal combustion engines, since for this type only is there a need for a possibility of extremely strict selection. Even this limited subject embraces about 16,000 patents! The subject is defined as follows: Apparatus for forming and making suitable for use in an internal combustion motor of a combustible mixture of: (a) air sucked in by the motor; (b) a liquid fuel or liquid fuels supplied by means of the subatmospheric pressure brought about by the motor, which fuel is mixed with the air by spraying (atomizing) it; (c) any added constituent. Fuel supply systems of the level control device of a carburettor are not dealt with as a part of the carburettor. Introduction to the scheme of characteristics First, this scheme is based on the point of view that very many objects, and in any event carburettors, can be defined by specifying: (a) characteristics which describe a functional part, constructional element, etc.; (b) characteristics which indicate the relation between such parts, elements, etc. In principle it is possible to define a machine, an apparatus, a construction, a structural (chemical) formula (3), and so on, by summation of their functional parts and indication of the relations between them. How far such a definition of objects can be realized in practice is of course a matter requiring separate study in each case. Furthermore, we took the point of view that in cases where it is important to be able to express variant embodiments of the characteristics, this must be effected in such a way that expression is also given in the code to the genus characteristic under which a species characteristic falls. If the above-mentioned conditions are fulfilled, the code for a given species

OCR for page 1071
--> of machine, apparatus, etc., includes the codes of all genus forms under which the species falls. Secondly, one can consider carburettors to be characterized by flow paths for various fluids (fuel, medium for the combustion of same, e.g., air, supplements, and mixtures of these), by regions of confluence or divergence of the flow paths (referred to henceforth as flow-path junctions), by devices situated in these flow paths (restricting devices, heating devices, mixing devices, etc.), by the means adapted for the control of the various functional elements, and by further relations between the fluid flow and the devices. The flow paths are characterized completely by the fluids for which they are intended. The flow-path junctions constitute a form of relation between the flow paths. The devices derive their raison d’être from the function they are intended to perform. Sometimes it is important merely that a device be present in order to perform a certain function; sometimes it is precisely the constructional details of a device that are of special importance. What is necessary, then, is characterization of the devices according to their function and, in so far as details that could be important are given, characterization of details. Now it is possible in many instances to ascribe more than one function even to a simple device. Thus a carburettor jet serves to form a mixture, but it can also serve to introduce extra fuel for starting purposes. However, the essential function of a jet as such, at any rate in a carburettor, with which this study is concerned, is to form a mixture, and the function during starting is rather an incidental use; the jet can be put to such a use, but it can equally well be made to perform other incidental functions. By characterization according to function is to be understood the essential function of the object as such in a carburettor (the characterization of incidental functions is also important, but this is kept separate from that of the device). Thirdly, with a view to coding, a system of 15 numbers is made use of for the scheme of characteristics. In order to be able to represent each number of the system by a single symbol, the numbers 10 to 15 are represented by letters, as follows: 10=t; 11=a; 12=b; 13=c; 14=d; 15=e. Fourthly, the scheme of characteristics comprises four sections: Main aspect Specification of detail of the devices Constituents of the flow Relationships; Control; Incidental functions

OCR for page 1071
--> A characteristic is represented by a single code symbol for the main aspect, by up to four code symbols for specification of the details, and by one code symbol for each flow constituent. In all cases only one main aspect is indicated and one aspect of detail, but all important constituents are indicated. To indicate an alternative aspect of detail the entire characteristic is repeated. A relationship, control, or incidental function is indicated by one code symbol for the main aspect, and up to four code symbols for the details. In addition, indication is given of the rows on which are recorded the characteristics to which the relationship, control, or incidental function applies. The last indication thus connects with the relationship, control, or incidental function, the characteristics with which the said relationship, etc., is associated. This connection will henceforth be termed the interrelation. We are of the opinion that the manner of coding indicated above only acquires practical significance for mechanized novelty search by the introduction of interrelation symbols. We have noticed in the meantime that such symbols are already in use by the Research group of the U.S. Patent Office (2, 5, 6). This group has made use of them for novelty searches relating to chemical substances. Such interrelation symbols are considered in greater detail under “Perforation of the interrelationships,” in the sections on “Punched card and machine” and “Interfixing.” Coding Scheme A Main aspect 1. Supply devices with special means for mixing one flow into another (e.g., fuel nozzles) 2. Shutdown or restriction devices, e.g., throttles (for float chambers see A=3) 3. Level control mechanisms 4. Transport devices 5. Heating devices (heaters) 6. Cooling devices 7. Devices for the improvement of the mixing 8. Strainers 9. Devices outside the carburettor; measuring devices t. Mixing devices b. Devices to split the constituents of a flow c. Carburettor d. Relations between more than one device and/or flow paths; control systems and incidental functions of devices e. Other devices or functions

OCR for page 1071
--> interfixing field at least those columns are perforated that correspond to rows in which the characteristics were found. As far as we know there are no machines commercially available that are able to conduct this method of selection. However, there are good grounds for believing that several commercially available sorting or collating machines can be adapted to our purposes without very great difficulty. In any case it appears that the Ilas Scanner (6) of the U.S. Patent Office completely fulfils the above requirements. Accordingly, it does not seem appropriate to go any further into the problem of the machines. It only remains to be mentioned that the development of the machines required is at present in a stage of preliminary discussion with the manufacturers. An initial test of the coding system can be done by using a standard collating machine, with a separate card for each characteristic. Such a test, covering about 1,000 patent specifications will probably be carried out a few months from now. THE SUBDIVISION OF THE CARD columns 1–8 spare “ 9 used for a perforation indicating the country in which the patent has been issued (punched in the usual way) “ 10–16 punched for indicating the patent number (usual way) “ 17–20 spare “ 21–40 punched row by row using 1–2–4–8 coding “ 21–24 main aspect “ 25–40 details of devices; nature of relationships, control and incidental functions “ 41–44 spare “ 45–60 punched row by row for indicating constituents; however, one column has been assigned to each constituent, in order to make it possible to record more than one constituent in each row “ 48–59 these also serve as interfixing field, this being possible because these interfixings are only punched in the same row as a relation, control or incidental function, and these are not coded for constituents. Here column 48 corresponds to row 12, column 49 to row 11, column 50 to row 0, column 51 to row 1, etc. “ 61–79 spare “ 80 coded for indicating additional cards A perforation in position 12 indicates that one or more succeeding cards refer to the same patent specification. A perforation in position 1 indicates the first of the additional cards, in position 2 the second, etc.

OCR for page 1071
--> THE CODE 1–2–4–8 AS APPLIED For ease of understanding, a table is appended showing this code, that is punched in four columns of one row; the perforations are indicated by a cross (x):   1 2 4 8 0   1 x   2   x     3 x x     4     x   5 x   x   6   x x   7 x x x   8       x 9 x     x t=10   x   x a=11 x x   x b=12     x x c=13 x   x x d=14   x x x e=15 x x x x Examples EXAMPLE I: GERMAN PATENT SPECIFICATION 833,883 (FIG. 1) In this specification a carburettor is described with a throttle valve in the form of a slide valve 3, positioned in the mixture flow path above the spraying device 7 and movable by means of a Bowden cable 4 (the air inlet of the carburettor is indicated at 1, the mixture discharge at 2). A control needle 6, reaching into the mouth of the fuel jet is joined to the slide valve 3. The air supply can be regulated by means of a hand-controlled choke 8, which is slidable in the throttle valve. The notch 12 on the choke cooperates with the throttle valve 3 in order to limit the lowermost position of the choke. This position lies more to the fully open position as the throttle valve is drawn upwards. This carburettor has thus an air flow path and a fuel flow path converging at a junction where a flow path originates for a mixture of air and fuel. The main jet lies in the junction. In the air flow path lies a continuously controllable slide valve, in the fuel flow path a continuously controllable needle valve and in the mixture (air with fuel) flow path a continuously controllable slide valve. The valves are built together. The needle valve is joined to the throttle valve so that these are coupled mechanically. By means of the notch 12 the throttle valve and the choke are also coupled mechanically.

OCR for page 1071
--> FIGURE 1

OCR for page 1071
--> These data are coded on a punched card as follows (Fig. 2). FIGURE 2 Row 9: characteristic I: a continuously controllable slide valve in an air flow path A=2 (=shutdown or restriction device) B=1–3–(1=with continuously controllable passage; – –3=slide valve) C=1 (1=air) Row 8: characteristic II: a continuously controllable needle valve in a fuel flow path A=2 B=1–2–(– –2=movable in the direction of the flow, e.g., needle valve) C=2 (2=fuel) Row 7: characteristic III: a continuously controllable slide valve in a flow path for a mixture with the constituents air and fuel A=2 B=1–3– C=both 1 and 2 Row 6: relation m: the flow paths according to the characteristics I, II, and III have a common junction, in which at least two flows converge; the main jet lies in this junction A d (d=relation) B=411–(4=junction of flow paths; 41=.at which two or more flows converge 411=.. the main jet lies in this junction) interfixing perforations relating to the rows 9, 8, and 7. Row 5: relation n: the devices according to the characteristics I, II, and III are built together and are also coupled mechanically A=d B 211 (2=coupling between devices; 21=. mechanical coupling; 2.1=. devices are built together) interfixing perforation relating to the rows 9, 8 and 7.

OCR for page 1071
--> Row 4: relation p: the devices according to the characteristics I, II, and III and the junction according to the relation m coincide at one spot A=d B=71– –(71=devices and junctions coincide at one spot) interfixing perforations relating to the rows 9, 8, 7, and 6. Row 3: incidental function t: the device according to characteristic I is a choke B=t111 (t1=incidental function of devices only t11=. starting, idling or compensating; t111=.. starting), interfixing perforation relating to row 9 EXAMPLE II: GERMAN PATENT SPECIFICATION 678,769 (FIG. 3) This invention concerns a carburettor in which a preliminary mixture of fuel and air is formed and by means of a spraying device delivered into the suction pipe of the motor. The invention provides means for keeping the preliminary FIGURE 3

OCR for page 1071
--> mixture under nearly constant pressure, independent of the fluctuation of the subatmospheric pressure in the suction pipe. In this manner the delivery of fuel is independent of the fluctuations of the pressure in the suction pipe. According to the invention there is an additional atmospheric air bleed to the space where the preliminary mixing is effected, which bleed is controlled by a valve dependent on the pressure in the suction pipe, in such a way that the opening of the valve increases with increasing difference in pressure between the atmosphere and the suction pipe. In the drawing there is a flow path (a) for atmospheric air with a butterfly valve (b). The fuel from the float chamber flows through tube (t) to the nozzle (i), where the fuel is delivered into the space (m). Opposite the nozzle (i) lies the end (r) of a slide valve (h). By way of an inlet (p) in the slide valve atmospheric air enters the part of space (m) lying under the slide valve. Thence the air flows upwards past the end (r) of the slide valve and the fuel nozzle, thereby becoming enriched with fuel to form the preliminary mixture. The distance between the end (r) of the slide valve and the opposite wall of the space (m) is controllable by movement of the slide valve. Moreover through a passage (n) additional atmospheric air may be sucked into space (m). This air supply is controlled by a needle valve fixed to the piston (v). The underside of the piston is subjected to atmospheric pressure, and its upper side to the pressure in pipe (k), which pressure depends on that in the intake pipe of the motor. An increase in the pressure difference between the atmosphere and the intake pipe of the motor results in an upward movement of the piston and thus in enlargement of the passage for additional air. On the other hand, a decrease in the pressure difference results in restriction of the passage for additional air. If the control device is properly proportioned the pressure in space (m) will remain nearly constant. The position of slide valve (h) is controlled by a cam on the axis of the butterfly valve (b). The slide valve (h) can be replaced by a needle valve controlling the opening of the fuel nozzle (i). The piston (v) can be replaced by a membrane. For this example more than 12 characteristics are needed, so it has to be coded on two cards. The interfixing facility is limited to data on one card, so if interfixing perforations are made in a card, this card has to contain the code of all characteristics to which those perforations pertain. For clarity it has been arranged that the first card of the example contains the complete code for the carburettor with slide valve (h) in the air inlet, and the second card contains the additional code necessary for recording the alternative construction in which the slide valve in the air inlet is replaced by a needle valve in the fuel inlet. This example (Fig. 4) is coded as in Table 1.

OCR for page 1071
--> Interfixing METHODS USED FOR INDICATING INTERRELATIONS In the examples of the preceding section the following relations are discernible between the partial codes or characteristics codes: FIGURE 4 1. Between a device coded under aspect B and its genus (as seen in the hierarchy of the code) 2a. Fuel and air both present in the same flow path (as seen by the indication of both constituents in one row, this being made possible by assigning a separate position in the row to each constituent) 2b. Indication for each device of the flow path in which it is located (assigning part of each row for coding one device) 3. The indication of the characteristics to which a relation or control pertains (by interfixing perforations indicating the fields—rows—in which these characteristics have been recorded)

OCR for page 1071
--> Methods 1 and 2 are often used for recording relations between codes, method 3 seldom. METHOD 1: CODE SHOWING THE HIERARCHY (3, 5) This method has the great advantage that in coding a species the genus and their relation is coded at the same time. This happens as it were automatically. If, however, the relation between genus and species is not inherent in the subject of the system, several genuses may have the same series of species. In the latter case it is preferable always to allot the same code to corresponding species; if, furthermore, these are given a fixed location method 2 is obtained. TABLE 1   Code Row no. Characteristic to be coded A B C Interfixing perforations pertaining to the rows   1 2 12 11 0 1 2 3 4 5 6 7 8 9   First card   9 flow path (t, i) for fuel . . . . .   2   8 a continuously controllable slide valve (h) in an air flow path (p; passage past r) 2 1 . 3 . 1   7 a continuously controllable valve (needle on piston v) movable in the direction of the flow in an air flow path (o, n) 2 1 . 2 . 1   6 a flow path (k) for air and fuel ending in a spraying device (e) 1 1 . . . 1+2   5 junction (m) of flow paths mentioned above d 4 1 . .   6 7 8 9 4 air flow path (a) with valve (b) rotatable on an axis perpendicular to flow path axis 2 1 . 1 . 1   3 flow path from e for air and fuel . . . . . 1+2   2 junction (e) of flow paths mentioned in the characteristics coded in the rows 6, 4, and 3; the main spraying device lies in this junction d 4 1 1 .   3 4   6   1 the devices mentioned in the characteristics coded in the rows 8 and 4 are coupled mechanically d 2 1 . .   4   8   0 the valve mentioned in the characteristic coded in row 7 is controlled in dependence on the pressure of the gases in the suction pipe of the motor d 9 2 4 .   7       Second card   (12 another card for the same patent follows a p e r f o r a t i o n in co lu m n 8 0) 9 a continuously controllable valve (s, i) in a fuel flow path (t, i) 2 1 . 2 .   2   8 a flow path (p″) for air . . . . . 1   7 same as for row 7 in the first card 2 1 . 2 . 1   6 same as for row 6 in the first card 1 1 . . . 1+2   5 junction (m) of flow paths mentioned above d 4 1 . .   6 7 8 9 4 same as for row 4 in the first card 2 1 . 1 . 1   3 the devices mentioned in the characteristics coded in the rows 8 and 4 are coupled mechanically d 2 1 . .   4   8   (1 this card is a first additional card a p e r f o r a t i o n i n r o w 1,       c o l u m n 8 0 )  

OCR for page 1071
--> METHOD 2: CODES IN WHICH SUBDIVISIONS ACCORDING TO DIFFERENT POINTS OF VIEW EACH HAVE A PREDETERMINED LOCATION IN A COMMON CODING FIELD This is a very attractive method, making it possible to build up a large variety of characteristics from a limited number of partial characteristics, in such a way that the characteristics can be inspected from all possible combinations of points of view. If there are many points of view the size of the field needed for coding a characteristic may become so large, that practical difficulties result. METHOD 3: INDICATION OF INTERRELATIONS BY SPECIAL PERFORATIONS (Compare refs. 5, 6, and 7) The most important relationships between the (partial) characteristics used are junction relations between flow paths and the locations of devices in a system of flow paths. However, many more relationships exist between devices (e.g., mode of coupling or building together) as well as between flow paths (e.g., their concentricity). It has not proved possible to record them all in a combined code by simple ordering of partial codes. One is therefore obliged either to abandon coding a considerable part of the interrelations existing between the characteristics, or to use special code indications that show which characteristics belong together. The interfixing perforations already mentioned constitute such special code indications. They make it possible, comparatively simply to record much more that can subsequently be retrieved than conventional coding methods. It becomes possible to restrict the codes to characteristics built up to a restricted length from partial characteristics, to record these in an arbitrary sequence and then indicate their interrelations by means of interfixing perforations (in fact this makes the practically usable code length greater by one dimension). WOULD THE SYSTEM NOT BE SUFFICIENTLY SELECTIVE EVEN WITHOUT INTERFIXINGS? So far it has always been supposed that interrelations present in the specifications must receive expression. Is this really necessary? Would not it be sufficient to make no attempt to indicate the interrelations but to restrict the coding to those relations that do not receive expression in the relative location of partial codes in a code series? It would still always be possible to find the structures containing the elements of the structure sought. Amongst these several will, of course, not have the interrelations essential to the structure sought, but this would be acceptable if it were only a low percentage. It is better to

OCR for page 1071
--> find too many documents than to overlook some. Are not the possibilities of selection already so sharp that it would still be acceptable to find, for example, twice the number of specifications that would be found making full use of selection on interrelations? Of the numerous theoretically possible interrelations in a code series, are not so few of practical value that leaving out the interrelations would not seriously harm the selectivity of the system? The final answer to this question can in this case only be given by comparative tests. However, it appears practically certain that the coding scheme for carburettors as explained in this paper would not be satisfactory without the indication of the interrelations. This is partly due to the fact that without the interfixing perforations the difference between many characteristics would disappear. This would be very evident in searching for a combination of characteristics in which the details of the separate component parts were of no importance. Every carburettor has an air flow path, a fuel flow path, a mixture flow path, and a junction of these paths. In the mixture flow path a control mechanism is usually present. Without interfixing, these characteristics are therefore in no way selective. A search instruction for the invention of Example I of the preceding section without the use of interfixing perforations would practically result in selection of carburettors with a choke and two component parts of any kind that are built together and mechanically coupled. Probably hundreds will meet these requirements. In Example II the selection would practically be only for carburettors having any component part controlled by subatmospheric pressure in the suction pipe. This is also very common. Finally it has been determined that the code of the first variant of the second example would be the same as that of 83 possible variants that are not described in the patent specification. Research group PROGRESS: TIME SPENT The research group is composed of the following personnel: The author of this report, for general supervision and guidance (additional job). Mr. J.Dekker, for code machine problems (additional job). He has specialized in the fields of classification and punched-card machines. Mr. F.J.Siegers, for guidance during the development of the code and supervision of the coding (additional job). He is an examining officer in the field of internal combustion engines. Mr. H.van der Esch, as technical assistant, who has also played an important part in the development of the code (virtually full-time job).

OCR for page 1071
--> Mr. A.A.Looyen, as technical assistant (half-time job). The development of a usable, unambiguous code has involved: the first three mentioned persons to the extent of 5 months, over-all time; the two technical assistants for 10 months, over-all; an experienced worker can code about 10 patents per day. REFERENCES 1. Report to the Secretary of Commerce by the Advisory Committee on Application of Machines to Patent Office Operations; Chairman V.Bush; Department of Commerce, Sales and Distribution Division, Commerce Building, Washington 25, D.C. U.S.A., December 22, 1954. 2. A.BLOEMEN. Classificatie in de organische chemie, Chemische Weekblad, vol. 37, no. 47 (1940), published by D.E.Centen’s Uitgevers Maatschappij, N.V., Amsterdam, pp. 619, 620. Abstracted in Chemical Abstracts, “Classification in organic chemistry,” p. 4799, 1942. 3. JAMES W.PERRY, ALLEN KENT, and MADELINE M.BERRY. Machine Literature Searching, Interscience Publishers, New York, 1956; especially “Generic encoding of index entries,” p. 12. U.S.A. Patent Office Research and Development Reports 4. SIMON M.NEWMAN. Problems in mechanizing the search in examining patent applications. 53310-U.S. Dept. of Comm-DC-1956, p. 28, “3. Creating and designing methods of, and apparatus for using, the precise and intricate patterns required.” 5. B.E.LANHAM, J.LEIBOWITZ, and H.R.KOLLER. Advances in mechanization of patent searching, Chemical Field. April 11, 1956. 6. Report No. 6., DON D.ANDREWS. Interrelated Logic Accumulating Scanner (Has), June 25, 1957. 7. Report No. 8. Recent Advances in Patent Office Searching: Steroid Compounds and Ilas, 1957. J.FROME, H.R.KOLLER, J.LEIBOWITZ, H.PFEFFER. Steroid Compounds. DON D.ANDREWS. Ilas. Patents 8. U.S.Patent 2,720,360. Card scanning mechanism, to H. P. LUHN; application date March 31, 1950; patented October 11, 1955. 9. French Patent 994,781, Procédé et machine pour classer et electionner les documents, to M.J.SAMAIN; application date May 15, 1944; patented August 14, 1955. With additions 55,907, 56,342, and 59,504, the last being almost equal to British Patent 662,819.