www.elsevier.com/locate/worpatin

World Patent Information 30 (2008) 5–20

English-language support tools for the use of Japanese F-termpatent subject searching online

Stephen Adams

Magister Ltd., Crown House, 231 Kings Road, Reading, Berkshire, RG1 4LS, UK

Abstract

The purpose of this paper is to provide an overview of the help tools available to the English-language patent searcher, to assist in thesystematic and informed selection and use of the F-term system. In order to show the effectiveness of these tools, some information on theofficial documentation, structure of the F-terms system and usefulness in searching is given. Information on both free-of-charge and com-mercial loads of the F-term data is reviewed. Some instances are highlighted where at present, support material is only available in Jap-anese, together with recommendations for expansion of English-language support.� 2007 Elsevier Ltd. All rights reserved.

Keywords: Japan; F-term; Subject searching; English version; PMGS; Notation; Patolis; Chemical Abstracts; PatBase

1. Background

The Japanese system of F-terms (File-forming Terms) isan internal subject indexing system applied to Japanesepatent documents. The system has its foundations in themove towards the ‘‘paperless patent office’’ in the early1980s [1] and was operational by 1987, reaching full func-tionality by 1994. The system is considerably more exten-sive than the International Patent Classification (IPC),consisting of some 340,000 sub-divisions, compared toapproximately 70,000 in the newly-released eighth editionof the IPC. It is still a relatively unknown search tool forthe Western patent searcher, and has only recently becomepublicly available in non-Japanese language systems. Thefirst English-language launch was via the Industrial Prop-erty Digital Library (IPDL) system at the Japan PatentOffice (JPO) website in 2001; other commercial implemen-tations have followed. Further background on the princi-ples of the system can be found in previous publicationsby Nakamura [2], Schellner [3], Ueno [4] and Nakata [5].

For searchers accustomed to using classification, it isworth pointing out that the philosophy of application ofF-terms is distinct from that of classification. The latter

0172-2190/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.

doi:10.1016/j.wpi.2007.05.008

E-mail address: stevea@magister.co.uk

seeks to classify the main inventive feature, whilst the for-mer indexes the technologies used in the invention. Assuch, F-terms provide a particularly useful tool for invalid-ity or freedom-to-operate searching, in which new applica-tions of known technology must be located.

F-term are only applied to Japanese documents, by per-sonnel from the Industrial Property Cooperation Center(IPCC), an independent body set up in 1985 and subse-quently designated by law as the official outsourced con-tractor for Japan Patent Office search work [6].

During the course of preparation of this paper, the CXPbeta-test site hosted by Paterra, which included a set ofF-term help material, was withdrawn, and it is not knownat this stage whether it will be re-introduced into the mainPaterra website. For the convenience of the reader, thecomments on the Paterra help material have been retainedin this article, as a way of illustrating different possiblemethods for supporting the F-term system.

2. Structure of the F-term deep-indexing system

2.1. Derivation from the FI classification

F-terms are multi-dimensional indexing codes, with atwo-part notation. The first part is a 5-character unit called

6 S. Adams / World Patent Information 30 (2008) 5–20

a ‘theme code’, which defines a broad technical field. Thesecond part is typically a 4-character unit, consisting of atwo-letter ‘viewpoint’ or ‘aspect’, followed by a two-digitextension, called the ‘figure’. The combination of view-pont + figure is referred to as a ‘term code’, although someauthors refer to this 4-character unit as ‘the F-term’ itself.For the purposes of this paper, the description ‘term code’will be used for the second part of the notation, and ‘F-term’ will be used to mean the complete 9-character string.A typical example of a complete F-term notation is:

4L037 CS03

in which the theme code is 4L037, the viewpoint is CS andthe full term code is CS03. In this particular example, thetheme code relates to the technical field of ‘inorganic fibres’,

G11B 7/00 to 7/0135 andG11B 7/28 to 7/30,ZTheme = 5D090

G11B 7/08 to 7/085,ZTheme = 5D117

G11B 7/09 to 7/10Theme = 5D118

G11B 7/12 to 7/22Theme = 5D789

G11B 7/24 to 7/24,572,ZTheme = 5D029

G11B 7/26 to G11B 7/26,531Theme = 5D121

G11B 7/00 to 7/0135 andG11B 7/28 to 7/30,ZTheme = 5D090

G11B 7/08 to 7/085,ZTheme = 5D117

G11B 7/09 to 7/10Theme = 5D118

G11B 7/12 to 7/22Theme = 5D789

G11B 7/24 to 7/24,572,ZTheme = 5D029

G11B 7/26 to G11B 7/26,531Theme = 5D121

Fig. 1. Distribution of group G11B 7/00 into themes.

5K

5K

5J1

5K

5K

5K

5K

5K004

5K029

H04J

H04K

H04L

5K

5K

5J1

5K

5K

5K

5K

5K004

5K029

H04J

H04K

H04L

Fig. 2. Distribution of FI sub-classes H04J,

the viewpoint identifies ‘chemical structure of the fibre’ andthe term code CS03 defines ‘carbon fibres’.

The number of documents which may be indexed by agiven theme code is relatively small; Takei [1] indicates thisis usually in the range of 3000–5000 (but this may haveincreased since his paper was written). A single documentwill typically carry many more F-terms than correspondingFI or IPC classification marks, providing a much moredetailed breakdown of the subject matter. Snapshot statis-tics from the end of 2006, supplied by Paterra, indicated anaverage of 14 F-terms per document, but individual caseswith 50 or more are not unknown.

The theme codes are constructed by recombining sub-divisions from the Japanese FI (File Index) classification.The FI is the internal enhanced International Patent Clas-sification (IPC) used by the Japan Patent Office (JPO). Thebasic structure of FI follows substantially the same hierar-chies as the IPC, but with finer divisions at the sub-grouplevel . A single theme code may group together a contigu-ous range of FI terms (for example, theme code 3G001 cov-ers the narrow range of FI sub-groups from F02D 41/00 toF02D 41/40 inclusive) or a non-contiguous set (for exam-ple, theme code 5K022 covers a wider range of groups fromFI sub-classes H04J and H04L). Some examples exist inwhich a theme code groups together intellectually-relatedmaterial from entirely different sections of the IPC/FI –theme code 5J104 (ciphering devices) draws from sub-clas-ses G09C, H04K and H04L. The converse is also true; therange of subject matter covered by sequential FI sub-groups may be allocated to different themes.

Figs. 1 and 2 illustrate these mapping relationshipsbetween FI and F-terms. In Fig. 1, the pie-chart representsthe FI sub-groups of main group G11B 7/00. Each slice of

022

028

04

104

059

043

047

5K030

5K031

5K032

5K033

5K034

022

028

04

104

059

043

047

5K030

5K031

5K032

5K033

5K034

H04K and H04L across F-term themes.

S. Adams / World Patent Information 30 (2008) 5–20 7

the pie-chart represents a range of sub-groups, and theboxes show the theme code into which each sub-grouprange is collated. Note that theme code 5D090 consists oftwo separate ranges, sub-groups G11B 7/00 to 7/0135inclusive and G11B 7/28 to 7/30,Z inclusive. In Fig. 2, ashort sequence of three FI sub-classes (H04J, H04K andH04L) is represented by the three pie-charts. Each pie-chart is divided according to the number of main groupsin the sub-class, and the distribution of subject matteracross the theme codes is shown by the arrows linking tothe boxes, each box representing one theme code. Note thatseveral theme codes draw from non-contiguous ranges ofFI groups; for example 5J104 is derived from the two maingroups (H04K 1/00 and H04K 3/00) of sub-class H04Kplus some subject matter from an H04L group, H04L 9/00. In a similar manner, the large slice from H04L repre-sents groups H04L 15/00 to 23/00 inclusive, all of whichcoalesce into a single theme code, 5K043, along with somematerial from H04L 5/00 (the remainder of which goes into5K022 or 5K028).

It is important to bear in mind that, although the F-termsystem is created by derivation from the FI, the two sys-tems are distinct and are developing in parallel. In themajority of cases (some 1800 themes out of a total of2600), the theme is elaborated into a set of term codes,forming the second part of the F-term notation. However,approximately 800 theme codes have not been elaborated,and remain as so-called ‘FI themes’, with no correspondingterm codes. Examples include theme 2H017 (covering FIsub-groups G01B 1/00 to 1/66), 2H061 (G03B 43/00 to43/02) and 3B012 (A45D 33/00 to 40/30). These ‘FIthemes’ can be used only as a broad technical filter. Inpractice, the JPO examiners have found that the FI systemprovides sufficient precision for searching purposes in thesesubject areas, so no development of the F-terms has takenplace. In the event that these technical fields become morecomplex or faster moving, and require the extra flexibilityof a full F-term system, then the term codes can be devel-oped within the existing theme. It is worth pointing outthat some searchers’ experiences suggest that FI andF-terms can be used in parallel within the same search,whilst others will always use exclusively F-terms if theyhave been elaborated, and only revert to FI if no F-termshave been elaborated for that area of technology. It isreported that some parts of the F-term system have beenabandoned in favour of FI [7] – a listing (in Japanese) ofaffected themes can be found at the JPO website [8]. Sincethe launch of the eighth edition of the IPC on 1st January2006, the Japanese FI system – originally based upon thefourth edition – has been modified to bring it more intocompliance with the reformed system.

2.2. Official documentation in English

The translation of the theme codes and their individualterm codes into English is now virtually complete. A sharedproject of the trilateral patent offices (European Patent

Office, JPO, United States Patent & Trademark Office)was completed in 1999; however, there have been subse-quent amendments to the system, and not all of these havebeen translated. It is estimated that some 20–30 themesand/or their corresponding term codes are amended orupdated per year, causing a temporary ‘hole’ in the Eng-lish-language version of the system [9].

The conventional manner in which the JPO displays theterm codes within a given theme code is in the form of amatrix structure, which is discussed further below. Someof the minor parts of these theme code matrices, such asshort header descriptions of update status, may remainincomplete, but the essential data to permit use of thetheme for searching is present.

In addition to the translation of the matrices of F-termterm codes, there is a limited amount of English-languagesupport documentation in the form of written explanationsof F-terms. These explanations which include coveragedetails, definitions and scope notes, effectively constitutethe ‘‘Manual for F-terms’’ and are created primarily forthe JPO examiners, to assist them to understand and searchF-terms. The process of translating these ‘F-term explana-tions’ into English is still proceeding. In mid-2006, the JPOannounced that a further 120 theme explanations had beenloaded onto their Patent Map Guidance System (PMGS)website [10], bringing the total to approximately 200 outof the 1700–1800 current theme codes. The theme explana-tions sometimes include information about the derivationof the terms used; in at least some instances, these havebeen effectively adopted without alteration from contempo-rary review articles in the field.

From the year 2000, the F-term theme codes and thecorresponding term codes began to be printed in full onthe front page of unexamined patent applications fromJapan (JP-A documents) [11]. Granted patents do not carrythem on the front page, although as noted below the elec-tronic records do, albeit with a substantial time-lag. Priorto 2000, a different form of notation appeared on JP-Adocuments, which included the theme group but not thefull term codes. As will be seen later, these facsimile recordscan be of help when trying to prepare an F-term basedstrategy. Searchers viewing electronic front pages shouldbe aware that the extensive listings of F-term term codescan often spill over onto the back page of the document.

2.3. Notation

The theme codes represent a partially-hierarchical divi-sion of subject matter, with the first two characters indica-tive of the JPO examining division which will handle thecase, and a further 3-character string which elaboratesthe field into a broad subject category. For example, themecodes with the prefix 5H represent the subject matter exam-ined by the control and electric power generation examin-ing division at the JPO. The first two characters of atheme code are referred to by some sources as a ‘themegroup’.

Table 1Top-level theme groups of the F-term system

2B 2C 2D 2E 2F 2G 2H 2K 2N3B 3C 3D 3E 3F 3G 3H 3J 3K 3L4B 4C 4D 4E 4F 4G 4H 4J 4K 4L 4M5B 5C 5D 5E 5F 5G 5H 5J 5K 5L 5M

Table 2Viewpoints within a theme code

Theme code Viewpoint Viewpoint title

4C076 (medicinalpreparations)

AA Forms

BB Location of drugadministration

CC Active ingredientsDD Inert ingredients (I)EE Inert ingredients (II)FF Purpose and functionsGG Methods of manufacture

8 S. Adams / World Patent Information 30 (2008) 5–20

At the time of writing (late 2006), the JPO’s website [12]shows a top-level hierarchy of 41 theme groups, as shownin Table 1.

However, the Patolis search guide [13] shows the samematrix with slight variation, consisting of an additionalgroup 9A. The 9A group links to a single apparent themecode of 9A001 but the corresponding matrix of term codesis not available in English. The author has been informedthat this 9A001 listing is used for searching computer soft-ware inventions in Japan, and is simply a term listing ratherthan a regular F-term theme. It has not been translatedinto English at this time.

The Paterra website [14] covered the same 41 groups asthe JPO, but noted that group 2N on the JPO site did notlink to a valid page, which is indeed true. The Patolis guidelinks group 2N to a single theme code 2N038 (books andspecial printed materials) corresponding to a range ofsub-groups from FI sub-class B42D. The JPO navigationtool for the FI system also shows a cross-reference fromB42D to theme 2N038. There is currently no hyperlink tomore detailed F-terms for this theme, so it would appearto be an FI theme only.

It is somewhat disconcerting to the user to discover dis-crepancies between sources at this gross level of the system.There may be perfectly valid reasons, perhaps to do withthe speed of loading revisions of the system on each web-site, but it does raise questions in the mind of the searcheras to which system can be trusted. It would be helpful if allthe support services could be more systematic in stating thecurrency of their data load, and the source of data used.

The next level elaborated, below the theme group nota-tion, is one or more theme codes. These are represented byfurther numerical extensions, not necessarily in a completesequence. For example, theme group 4K (metallurgy) con-sists of 61 theme codes ranging from 4K001 to 4K006,4K011 to 4K063, 4K070 and 4K501. Group 3F has 108corresponding themes, with large gaps in the sequence:3F011 to 3F064, then 3F068, 3F070, 3F072, 3F074 to3F081, then 3F100 to 3F115 inclusive, and so on. Gener-ally speaking, the third character of a theme code is 0 fora code covering well-established or slower-moving technol-ogy, whilst newly-established themes created to accommo-date more rapidly-changing areas will start with a 1, 2 orhigher digit. However, these changes are not a systematic‘edition’ indicator, merely an approximate guide to theage of the theme.

The second distinct part of a full F-term notation is the‘term code’, consisting of a two-letter viewpoint and (usu-ally) a two-digit figure. The viewpoint is a primary sub-

division of the subject matter in the theme code; there aretypically 5 to 15 viewpoints per theme. Note that the view-points must always be read in the context of the technolog-ical area of the theme code, not just its title; for example,theme code 4C076 is entitled ‘Medicinal preparations’ butis derived from FI groups A61K 9/00-9/72 and A61K 47/00-47/48, which are for medicinal preparations character-ised by special physical form and non-active ingredients

used; consequently, the whole content of the theme isbiased in this direction, and viewpoint FF which is entitled‘Purpose and Functions’ relates only to the purpose andfunctions of such ingredients or forms, not the purpose(therapeutic use) of the drugs themselves. Table 2 showsall the viewpoints associated with theme code 4C076. Anindexer who decides to apply F-terms from a given themecode may choose from more than one viewpoint, and applyas many individual term codes from each viewpoint asseem appropriate to the subject matter. This contrasts withthe principles governing IPC, which normally encouragethe classifier to apply only the single most appropriatesub-group from within a main group listing.

Further sub-divisions within the viewpoint provide moredetailed individual term codes. Although the overall nota-tion generally remains the same length, there are dot levelhierarchies within the definitions, just as for IPC and FIclassification, as shown in Table 3.

The individual term codes of a theme are arranged bythe JPO into a matrix layout, in which a viewpoint occupiesone or more rows, and the hierarchical level of the termcodes reads across the rows from left to right, sometimeswrapping round to the next row. This means that it isimportant for the user to ensure that they have browsedthe entire hierarchy of terms before selecting the term codesto use. Part of an example matrix of term codes, for themecode 5F041 (Light-emitting diodes), is shown in Fig. 3 – thedefinitions of term codes have been omitted in order to

Table 3Hierarchical levels within a theme code

Notation Level Hierarchical term definition

4 Chemistry4D Theme group Chemical engineering4D002 Theme code Processing of waste gases4D002 AA Viewpoint Components to be processed. . .. Figure4D002 AA16 . Nitriles4D002 AA17 .Halogens and halides4D002 AA18 . .Chlorine (Cl2) and chlorides4D002 AA19 . . .Hydrochloric acid (HCl)4D002 AA21 . . .Organic chlorine compounds4D002 AA22 . . Fluorine (F2) and fluorides4D002 AA23 . . .HF4D002 AA24 . .Bromine (Br2) and bromides4D002 AA25 . .Iodine (I2) and iodides4D002 AA26 . Silicon (Si) compounds

S. Adams / World Patent Information 30 (2008) 5–20 9

emphasise the structure of the viewpoint rows. In thisexample, each viewpoint is represented by a single two-let-ter notation, with the exception of the ‘Forms of LED’viewpoint, which stretches across the CA and CB notation.The F-term matrices, as published by the JPO, occupy amaximum of 10 columns, such as AA01 to AA10, beforestarting a new row with term AA11 and so on. The overlaponto the next row appears to have no significance in rela-tion to the hierarchical level of the term; for exampleAA11 is a two-dot level term, as is AA10 in the row above,but AA21 is a one-dot level, below the ‘main group’ AA00,and not related to its predecessor term AA14.

Note that, just as with the theme codes, there are breaksin the numerical sequence of the term codes; see for exam-

Viewpoint title and notation

Purpose (AA) AA01 AA02 AA03

AA11 AA12 AA13

AA21 AA22 AA23

AA31 AA32 AA33

AA41 AA42 AA43

Drive circuits (BB) BB01 BB02 BB03

BB11 BB12 BB13

BB21 BB22 BB23

BB31 BB32 BB33

Forms of LED (I) (CA) CA01 CA02 CA03

CA11 CA12 CA13

CA21 CA22 CA23

CA31 CA32 CA33

CA41 CA42 CA43

CA51 CA52 CA53

CA61 CA62 CA63

CA71 CA72 CA73

CA81 CA82 CA83

CA91 CA92 CA93

Forms of LED (II) (CB) CB01 CB02 CB03

CB11 CB12 CB13

CB21 CB22 CB23

CB31 CB32 CB33

Shapes of finished products (DB) DB01 DB02 DB03

………………… …… …… ……

Applications (FF) FF01 FF02 FF03

FF11 FF12 FF13

Fig. 3. Schematic view of part of F-term matrix for theme code 5

ple the absence of term code AA20 from the sequence intheme code 4D002 (Table 3), and missing term codesCA09, CA59 and others in the matrix of 5F041 (Fig. 3).The hierarchical nature of the term codes can be exploitedautomatically in some search engines – see below for morediscussion on this point.

In the absence of their parent theme code, the notationfor the term codes is non-unique and cannot be used in iso-lation: for example, term code AA23 in theme 4D002 (pro-cessing of waste gases) refers to the processing of hydrogenfluoride, whereas the same term code AA23 in theme4D004 (processing of solid wastes) refers to disposal ofscrap batteries.

It was noted above that the term code is normally4-characters. However, there is provision for some termcodes to be extended beyond this, by the addition of fur-ther letters or symbols (sometimes referred to as ‘additivecodes’). These extensions are not used in all theme codes.Statistics from late 2006, supplied by Paterra, suggest thataround 8% of any given week’s Kokai documents can bearF-terms with these extensions. The definitions are not listeddirectly on the theme code matrices at the JPO website, butthey are present as a link from the viewpoint code in theleft-hand column of the matrix. For example, from thematrix view for theme code 4C084, clicking on the view-point code ZC leads to a page headed ‘‘F-term additionalcode list’’, and clicking on theme 4C084 from this list leadsto the additional codes available for each term code in thetheme. Some examples are shown in Table 4.

Note that some subject matter extensions can be indi-cated by the addition of chemical symbols rather than sim-

Individual F-term term codes

AA04 AA05 AA06 AA07 AA08 AA09 AA10

AA14

AA24 AA25

AA34 AA35 AA36 AA37 AA38 AA39 AA40

AA44 AA45 AA46 AA47

BB04 BB05 BB06 BB07 BB08 BB09 BB10

BB24 BB25 BB26 BB27

BB34

CA04 CA05 CA06 CA07 CA08 CA10

CA14

CA24 CA25

CA34 CA35 CA36 CA37 CA38 CA39 CA40

CA44 CA45 CA46 CA47 CA48 CA49 CA50

CA54 CA55 CA56 CA57 CA58 CA60

CA64 CA65 CA66 CA67

CA74 CA75 CA76 CA77

CA84 CA85 CA86 CA87 CA88

CA94 CA98 CA99

CB04 CB05 CB06

CB14 CB15 CB16

CB24 CB25 CB27 CB28 CB29

CB36

DB04 DB05 DB06 DB07 DB08 DB09

…… …… …… …… …… …… ……

FF04 FF05 FF06

FF14 FF15 FF16

F041 (Light-emitting diodes), showing viewpoints along rows.

Table 4Some example of F-term additive code extensions

Theme code Viewpoint Exampleextension

Definition

2H091 Liquid Crystals FA Optical elements X Optical element sited in front of liquid crystal3E067 Packaging BA Types of unit container B Inner container4B065 Microorganisms and culturing

processesAA Types of microorganisms Y Microorganism used as genetic source for

recombinant4C076 Medicinal preparations DD Inert ingredients T Flavouring substances4C083 Cosmetics AA Natural or incompletely defined

ingredients1 Present in claims

4C084 Drugs containing proteins orpeptides

ZC Medical uses 2 Component of mixture

10 S. Adams / World Patent Information 30 (2008) 5–20

ple letters or numbers. The example in Table 4 from theme4C083 shows that the notation can also be used to definethe part of the document in which the subject matter beingindexed occurs, such as claims or description. However, theadoption of this latter approach seems to be inconsistent;there are also instances in which the basic 4-character termcodes incorporate such document segment information: forexample, viewpoint AA of F-terms 4C084 relates to typesof inventions, and term code AA02 is for ‘‘claims’’.

As noted below, PatBase seems to be the only commer-cial system which has loaded the definitions of the exten-sions in English; some shorter definitions are included inthe PatBase system itself, whilst others are available as alink from the PatBase viewpoint code into the JPO’sPMGS.

3. Back-file availability, currency of F-terms

One of the challenges in learning how to use F-terms liesin determining the range of documents which can beretrieved with the system. Unfortunately, the help files atthe JPO website combine the data coverage tables for F-terms with those for the related FI system, so it is notimmediately clear whether the coverage is the same forboth systems. Personal communication [7] confirms thatthe entire back-file has been indexed. The JPO website helpmessages are found under the ‘Stored data’ button for ‘FI/F-term search’. The earliest searchable data described as‘examined patent applications’ are from 1914 (the earliestyear when they were published) and from 1971 for unexam-ined cases (i.e. the start of the deferred examination systemin Japan). There is a separate undated componentdescribed as ‘patent specifications’ (from No. 1 to777764) which denotes the earlier records back to 1885.Statements on the Paterra website also indicated that F-terms have been retrospectively applied back to 1885 (thebeginning of the modern Japanese patent system). How-ever, a number of the public search systems have onlyloaded a smaller proportion of the data (for example,Chemical Abstracts coverage commences with documentspublished from January 2004 only, and PatBase goes backto the mid-1980s). Clearly, it is important from the search-er’s point of view to know the year range which may beretrieved with this system.

Given the complexity of the system, it would be under-standable for there to be a time-lag between publicationdate of a document and availability of the F-terms in elec-tronic form for searching. However, this lag seems to varyaccording to different document kinds. The JPO help filesas of December 2006 were indicating that the English-lan-guage system had F-terms for JP-A and JP-U documentsloaded up to late November (i.e. less than four weeksdelay). The corresponding Japanese-language file typicallyhas a currency of less than one week. However, the appli-cation of F-terms to granted patents (JP-B, Toroku) wasonly current up to September 2004. In the course of prepar-ing this paper, other tests indicated that the JP-B termswere in fact more current than this – it is possible that thereis a mismatch between the help files and the actual data-base, or that F-terms are loaded for the B documents in lar-ger batches, causing intermittent timeliness problems.

A further issue relates to the amendment of the F-termsystem. Other authors have indicated that are systematicrevisions to the system, but provide no information onthe process for re-indexing the back-file whenever a termis subdivided or replaced (the present author understandsthat this information is available in Japanese to users ofthe local IPDL version, but not currently in English). Thisplaces the searcher in a dilemma similar to that with theECLA system, where it can be very difficult to determinewhen a group of documents (or rather, the electronic bib-liographic records thereof) has been changed to takeaccount of class revisions.

4. Strategies for locating relevant terms in an unfamiliar

classification or indexing system

The author’s experience in training searchers in a rangeof classification and indexing systems has shown that thereare essentially four methods which they will adopt in orderto identify relevant marks.

These methods are:

(a) using a catchword index or similar intellectually-pre-pared ‘back-of-book’-style tool to look up a word orphrase and be referred to a suitable entry point,

(b) direct searching of the text elements of the classifica-tion system,

S. Adams / World Patent Information 30 (2008) 5–20 11

(c) using one or more documents of known relevance toidentify candidate terms, which are then used as theentry point to the system definitions,

(d) browsing the structure of the system from first princi-ples, usually through hierarchical sub-divisions (suchas the IPC section to class to sub-class etc.)

Method (a) works well with the IPC, using either theofficial Catchword Index or the German Stich- und

Schlagworterverzeichnis, or for the US patent classificationusing the Index. Unfortunately, no such index exists inEnglish for the F-term system, although Paterra did pro-vide some word-based navigation aids at the top-level ofthe system (as opposed to a search engine for using method(b)).

Method (b) is a useful option in some circumstances,and can be deployed in some electronic files such as theECLADEF file on Questel-Orbit (for ECLA) or thePATIPC file on STN (for the various IPC editions). How-ever, its accuracy usually depends upon whether the textstring being searched for is utilised within the somewhatformal descriptions of the schedules. For example, asearcher hoping to locate the relevant IPC group for ‘pyr-idine compounds’ by this method may well discover groupC07D 211/00 ‘‘Heterocyclic compounds containing hydro-

genated pyridine rings, not condensed with other rings’’,which is fairly close, but would fail to locate directly themore accurate answer, which is group C07D 213/00, enti-tled ‘‘Heterocyclic compounds containing six-memberedrings, not condensed with other rings, with one nitrogenatom as the only ring hetero atom and three or more dou-ble bonds between ring members or between ring membersand non-ring members’’. The latter group is a better choicefor the topic of interest, but its title does not use the word‘pyridine’ at all. It is important for the searcher to knowwhether their search engine is addressing the originalschedules (as in ECLADEF) or using a modified strategyto identify and rank terms (such as WIPO’s TACSY ‘nat-ural language’ search of the IPC, or the esp@cenet ECLAsearch utility, which uses titles and abstracts from classifieddocuments). With regard to F-terms, Patolis provides asearch engine which operates on the text of the F-term termcodes, but the JPO website does not. The Paterra websiteoffered a similar search engine to the Patolis one, beforeit closed down.

Method (c) is by far the most popular for newcomers tothe IPC and ECLA, and it can be used for F-term withsome degree of confidence. However, its usefulness is lim-ited by whether the searcher can view up-to-date biblio-graphic records. There are two reasons for this. Firstly, asample search could typically identify a range of JP-A doc-uments spread across several years, but F-terms will onlyappear on the facsimile front page of cases published afterthe year 2000; any earlier documents, which may be back-indexed by additional appropriate F-terms, will not showany notation on their front pages at all. In order to gainan accurate picture of all the possible F-terms which may

be relevant, the searcher needs to be able to view a currentelectronic version of the bibliographic data, not rely upon afacsimile image of a document or static bibliographicrecords. Secondly, since the F-terms are a dynamic system,even the terms printed on newer front pages may not be anaccurate record of the current terms applied to that docu-ment. For these reasons, this method is only a reliableguide to identifying suitable search terms when the searchercan view full electronic bibliographic records. This optionis not available from the JPO website in English, but Jap-anese-speaking users can view current status of termsapplied to an individual record – see section below. Usersof the now-defunct English-language version of Patoliswere able to check these records, as are users of PatBaseand Chemical Abstracts Service on STN. Users also needto be aware whether or not the electronic records areupdated to take account of F-term revisions; Minesoft haveindicated that they do reload F-terms periodically into Pat-Base, but Chemical Abstracts is not currently doing so.

The final method for the new user of F-terms is togain a familiarity with the structure of the F-term systemand browse down from first principles to find suitableterms (method (d)). This has the advantage that the pro-cess of navigation encourages the user to examine a rangeof levels of detail, and they may find themselves being re-directed to more suitable areas of technology. The noviceuser thus gains some experience of the whole system,whereas some of the other methods will direct them tocandidate terms in the middle of the system, leading toa risk that they may select terms out of context, or beunaware of alternative, better terms for their search.Method (d), or some variant of it, is available on a widerange of search systems, which will be discussed in moredetail below.

5. Sources of help on F-terms

The current English-language loads of F-terms are lim-ited to the official JPO website, the Patolis search guide,Chemical Abstracts Service and PatBase from MinesoftLtd. Discussion of the features on the Paterra website arealso included in this section.

At the time of writing, none of the Thomson patentinformation products (World Patent Index, Delphion,MicroPatent) has loaded F-terms. MicroPatent hasincluded ECLA in some records, and indicated an interestin including F-terms, but there is no tangible record of pro-gress so far. Similarly, World Patent Index has recentlyincluded some US classification material, in the new recordformat, but has not indicated any clear intention to includeother systems.

5.1. Patent Map Guidance System (PMGS)/JPO

The PMGS is a free-of-charge part of the Japan PatentOffice website, available under the same menu as the othersearching services and managed by the National Center for

12 S. Adams / World Patent Information 30 (2008) 5–20

Industrial Property Information and Training (NCIPI,now referred to by the abbreviation INPIT) <http://www.ipdl.ncipi.go.jp/homepg_e.ipdl>. The PMGS is a linkfrom this home page, to the address <http://www5.ipdl.ncipi.go.jp/pmgs1/pmgs1/pmgs_E>.

The PMGS uses a common interface for searching andnavigating both the FI classification scheme and the F-terms. A screensnap of the interface is shown at Fig. 4.When searching F-terms, the results display will differaccording to which of the two ‘radio button’ options isselected (right-hand highlight box). With the ‘F-term list’option selected, the user may input either a theme groupor a theme code into the search window. If they search ona theme group, they are then presented with a list of cor-responding theme codes; if the user inputs a theme code,they are taken directly to the F-term matrix. The otheroption, ‘F-term description’, is largely ineffective for theEnglish interface, since much of the information has notyet been translated – as noted above, only some 200 outof 1800 theme codes have these supplementary descrip-tions in English at present.

In addition to a direct search, the word ‘F-term’ to theleft of the search box (left-hand highlight box in Fig. 4)is a hyperlink, which takes the user initially to the hierarchyof 42 theme groups shown in Table 1 above.

Neither search provides direct access to the text wordsof the theme code titles, viewpoints or term codes. This isunfortunate, since the novice user has no guidance con-cerning the meaning of the theme groups. Even if they

Fig. 4. Home page of JPO’s Pa

can find a way to select the correct group, they are thenpresented with a long list of the theme codes themselves,with minimal or short titles which do not provide muchguidance to selecting the correct technical area.

The alternative, recommended, approach to navigatingto a theme group is by means of the FI schedules. Asearcher familiar with a given portion of the FI can use thatknowledge to browse the FI groups using the PMGS, fromthe same starting point in Fig. 4. The display of FI groupsthen provides a link to the theme code(s) corresponding tothat group. Clicking this link takes the user to the fullmatrix of term codes for that theme code. An example isshown in Fig. 5, in which the user has browsed down theFI system to group A01G 1/00 – the links on the right-hand side of the page show that this subject matter hasbeen put into themes 2B011 and 2B022. Clicking on thelink of either theme code will take the user to the corre-sponding matrix of F-term term codes.

As indicated above, there is a method whereby Japa-nese-speaking users of the IPDL can view the current F-terms applied to a document. The EPO has created a helpfile for navigating to these records, as part of the FAQsection on Japanese information [15]. In order to work cor-rectly, the searcher needs a Kokai (JP-A) publication num-ber. The direct URL for the document is <http://patentinfo.european-patent-office.org/prod_serv/far_east/japan/db_search/_pdf/finding_fi_classes_f_terms.pdf>.

The PMGS is a separate module from the actual F-terms search engine on the JPO website, which is found

tent Map Guidance System.

Fig. 5. Navigating to the F-terms via FI classification.

S. Adams / World Patent Information 30 (2008) 5–20 13

off the NCIPI home page at ‘FI/F-terms search’ or thedirect URL <http://www4.ipdl.ncipi.go.jp/Tokujitu/tjf-termena.ipdl?N0000=114>. One little-known aspect ofthe search engine is the ability to control the hierarchicallevel of the F-terms retrieved. This is achieved by addinga prefix to the desired term code. If a search strategyincludes a $ prefix, then only documents with that termcode will be retrieved. By default, without the $ symbol,the system automatically retrieves both the search termand all term codes below it in the hierarchy. From theexample hierarchy in Table 3 above, searching with theterm $AA18 will retrieve only documents coded with4D002AA18 itself, whereas the default term without the$ prefix will retrieve documents coded with 4D002AA18,AA19 or AA21.

The JPO search engine also provides a means to searchthe term code extensions, referred to above in the sectionon Notation. In the printed format, there is no additionalspacing between the normal 4-character term code andany extension; for example, the term code CA09 (sub-strate materials) in theme 5F032 (element separation,manufacture of solid-state devices) can be refined by theaddition of molecular symbols to indicate specific sub-strates, hence CA09SiO for a silicon oxide substrate. Inorder for this to run on the JPO search engine, a dot (.)must be inserted between the fourth and fifth characters,thus 5F032CA09.SiO. Further information on the syntaxfor searching term code extensions is provided on theJPO website. As noted above, the definitions are availablevia a link from the viewpoint code in the appropriateF-term matrix.

5.2. Patolis-e Search Guide/Patolis Corp

As indicated above, the full Patolis-e search serviceclosed, for direct access, during September 2006. Part ofthe company’s website was dedicated to support tools for anumber of different indexing and classification systems,including F-terms, and this has been kept open for the pres-ent. However, users should be aware that the timeliness ofupdating for this ancillary information may not be asprompt as other sources; as always, it is wise to check in morethan one place for information, before using it in a search.

The Search Guide is available at <http://search.p4.pato-lis.co.jp/search_en.html>, and is shown at Fig. 6. TheGuide has three frames – the left-hand column consists ofthe theme groups from Table 1, each link of which opensa listing of the corresponding theme codes in the main centreframe. The top frame is a search engine, split into two-parts.

The bottom ‘F-term’ window of the search engineenables a user to input a theme code and retrieve the cor-responding matrix, or to input a complete F-term (themecode + term code, no spaces) and retrieve a tabular outputshowing the hierarchical position of the search term in itsproper context. This facility is rather more user-friendlythan the corresponding JPO search window, in that the list-ing of theme groups is retained throughout the process inthe left-hand frame. This makes it easier to navigate fromone theme code to another without backing up to thetheme group listing or the search window each time. Thisspeeds up the process of checking definitions of a largenumber of F-terms from a set of known relevantdocuments.

Fig. 6. Home page of Patolis help system.

14 S. Adams / World Patent Information 30 (2008) 5–20

The top ‘Keyword’ search window in the top frameintroduces a new facility, unavailable on the JPO website.This is the ability to directly search words from the textof the F-terms, either at the level of the theme code titlesor the text of each individual term code. In order toincrease precision, it is possible to specify both a keywordand an F-term theme/group, by adding a second term intothe bottom search window – there is a fixed Boolean‘AND’ operator between the two.

This search facility can make it much easier for anew user of F-terms to start the process of locatingappropriate terms. For example, a search on the key-word ‘rice’, limited to the theme code titles only, pro-duces 5 hits, of which 4 are directly relevant; 2B001(rice transplanter frames), 2B063 (soil-engagement ele-ments of rice-planting machines), 2B065 (planting unitsof rice-transplanting machines) and 2B396 (processingof unhulled rice in threshing machines). The fifth hitindicates that this search engine could produce someunexpected results; the hit is on theme 2H092 (part ofthe liquid crystal theme codes, relating to electrodesand active matrices).

By specifying ‘all’ instead of ‘Title of the theme codesonly’, the user can retrieve one or more individual F-terms.

Table 5Display format of F-term entry from Patolis search engine

Theme Title of the theme F-term

4B018 Coloring foods and improving nutritive qualities 4B018M4B018M4B018M4B018M4B018M

The hit list is presented in the form of a table, with the fullF-term (9-characters, no spaces) available as a link and thehit term highlighted in red within the text of the term. Forindented terms, the dot level is shown. Clicking on the indi-vidual F-term within the hit list takes the user to a second dis-play showing the hierarchical position of the hit term. Afurther link from here takes the user to the theme codematrix.

By way of example, the search term ‘shiitake’ produces ahit list of two F-terms: 2B011AA04 and 4B018MD83. Thelatter shows the word as a 4-dot level term. Clicking on thisrecord produces a display (shown in Table 5) which showsthe context of the term more explicitly.

Note that the description in upper-case opposite theMD00 term is the title for viewpoint MD, which sets thecontext for all lower terms. A useful feature of this display,in contrast to viewing the entire matrix, is that intermediateterms at the same hierarchical level are omitted, making iteasier for the user to understand the structure of the terms.Thus, term 4B018MD81 (Yeast), which is another three-dot level term appearing between MD80 and MD82 inthe matrix, has been removed.

As with any search engine of this type, its accuracy inretrieving relevant F-terms is dependent upon the search

F-term description

D00 NUTRITION MODIFYING SUBSTANCES AND FUNGID79 . Microorganisms, originating from microorganismsD80 . . FungiD82 . . . Mushroom (Basidiomycete)D83 . . . . Shiitake

S. Adams / World Patent Information 30 (2008) 5–20 15

words being present in the text itself (see discussion ofmethod (b) above). Notice that the translation into Englishof the above example uses American spelling (‘Coloringfoods. . .’). This appears to have been used consistently (asearch on ‘sulphur’ in the theme code titles yielded zerohits, but ‘sulfur’ produced 6). Since the entire system hasbeen translated from Japanese, it is likely that there willalso be some instances where technical terms have beenrendered in unconventional or non-idiomatic English.

The search engine appears to treat two-word entries as aBoolean ‘AND’, although this will effectively retrieve thecorresponding phrase; hence, a search for ‘isotropic etch-ing’ (without punctuation) retrieved 20 F-terms (of whichabout half were found because of the word ‘anisotropic’),whilst a search for the two-words ‘catalyst’ and ‘inhibitor’entered without an operator between them produced hitswith both terms separated by one or more additionalwords. No explicit operators are provided for – if a termsuch as ‘AND’ or ‘OR’ is included in the search window,these words are treated as if they are a search term.

Nonetheless, even with these limitations, the Patolissearch engine provides a powerful means to enter the F-term system. The user should, of course, always ensure thatthey check the context of a term before using it in a search,in case higher-level terms change the implicit meaning.

5.3. Guide to F-terms/Paterra Inc.(withdrawn)

Paterra, Inc. is best known for the InstantMT� service,providing machine translations of Japanese patent docu-mentation. A sub-site of their website, at <http://cxp.pater-ra.com>, provided a range of explanatory material to theF-term system. Certain parts of the data were unique,based upon Paterra’s own translation work.

The Paterra site, as with the Patolis one, provided twomechanisms to approach F-terms; navigation down thehierarchy or direct search. As noted above, the same setof 41 theme groups was provided as a first level (but withtitles to help selection, which are not available from theJPO). Clicking on any theme group led to a listing of thecorresponding theme codes, again with titles, which couldbe used to connect to the individual theme code matrices.

The direct search option appeared to function in muchthe same way as the Patolis guide, presenting the user witha hit list in a KWIC-type format, showing the theme codeand individual figure.

In addition to these two approaches, Paterra had alsoloaded a list of 76 so-called super-categories, each of whichlinked to a list of theme codes in that subject area. Thesesuper-categories are used in the original Japanese versionsof the F-term matrices, but have not been translated in thePGMS English version. Within each super-category, thetheme codes may be from widely differing theme groups;for example, ‘Electronic Games’ leads to theme codes fromgroups 2C, 5C and 5G. As before, a user could link froman individual theme code to its corresponding matrix.The super-category approach was an interesting intermedi-

ate between, at one extreme, navigating ab initio fromtheme group or theme code level down to the individualterm and, at the other, searching the text of individual termcodes.

5.4. Chemical Abstracts/STN International

The commercial load of the Chemical Abstracts filefrom Chemical Abstracts Service (CAS�) on the STNInternational host has recently included some F-term data.At the time of writing, F-terms have been added to recordsfor Japanese patents published after 1st January 2004, andthere is no indication that they will be added for recordsbefore this. This is unfortunate, since much of the powerof the F-term system lies in the fact that it can be used tofor retrieval across many year’s-worth of data. Japanesepatents are covered within the CAS system since 1918 inthe chemical technology area, and the JPO coverage tablesindicate that back-files exist with F-term data to this date.The CAS selection policy would eliminate any purely elec-trical or mechanical patents from coverage, but large num-bers in fields such as electronics are covered, when there issufficient chemical information to merit inclusion.

One feature of the STN system is the various online the-sauri, designed to help the user to identify search terms.The F-terms have been loaded as a thesaurus, and can beviewed and navigated within the system without connectcost, using the ZCAPlus file. It was announced at the endof 2006 that this thesaurus had been enhanced with multi-level Narrower Terms (NT), enabling the types of displaydescribed below. However, the data for this thesaurus werenot sourced from the JPO, and although it includes non-chemical areas and possibly some terms which have notbeen cited in the current range of Japanese documents cov-ered by CAS, it is probably not as complete a listing asfound in the PMGS or Patolis.

The syntax for searching and displaying F-terms in theCA/CAPlus files is slightly different to that seen in theJPO and similar sites. A stroke symbol (/) is placed betweenthe theme code and the term code, for example 4C076/BB11.

The F-terms have been loaded in a special field(/FTERM) in the file, which is available for use with theANALYZE command. This provides a means to extractF-terms from the records in a hit set and rank them inorder of frequency of occurrence. Consequently, it is possi-ble for a searcher to conduct an initial search using text,classifications or other search terms, and then examinethe F-terms which have been assigned to the results(method (c) above). Display using a standard bibliographicformat enables the user to see which terms have been asso-ciated with a single record; alternatively, the ANALYZEcommand enables the user to view an aggregated rankingof all terms assigned to all records in the list. The frequencylisting gives some guidance to the most commonly-appliedF-terms in that subject area. Before using a term, however,it is strongly advised that the searcher checks the full theme

16 S. Adams / World Patent Information 30 (2008) 5–20

code matrix for each candidate term, in order to under-stand the context of the term and other related term codes.

By using the thesaurus, the user is also able to browsethe hierarchy either side of a selected F-terms, in order togain insight into the context and surrounding terms. How-ever, there is no direct way in which a user can search thetext titles of individual F-term term codes or theme codes,so this structure is limited in its effectiveness as a toolenabling a new user to identify possibly relevant terms.

The thesaurus command on STN has a number of differ-ent parameters, which affect the manner of display of thethesaurus. The parameter is indicated by a suffix to theterm. Using the syntax EXPAND [F-term]+ALL/FTERMdisplays the chosen term and all its narrower terms, endingat the next term from the same hierarchical level as the cho-sen one. For example, viewing the thesaurus around F-term 4C076/AA36 (a one-dot level below the viewpointAA) yields the display at Table 6. The STN version showsall the indents (including the theme group, theme code andviewpoint levels) using a dot structure, which can be some-what confusing to someone familiar with the JPO matrices.The selected term is shown as a 5-dot level; its narrower

Table 6STN thesaurus of F-terms (HIE display)

E1 0 BT5

E2 0 BT4

E3 57728 BT3

E4 8127 BT2

E5 4 BT1

E6 1039 –>

E7 251 NT1

E8 194 NT1

E9 52 NT2

E10 87 NT3

E11 111 NT3

E12 105 NT1

E13 56 NT2

E14 291 NT2

E15 116 NT3

E16 6 NT1

E17 17 NT1

E18 20 NT1

E19 61 NT1

terms are codes AA37 to AA49 inclusive, whilst the nextterm at the same level as AA36 is AA51, where the displayterminates. The advantage of this display is that the noviceuser is shown the hierarchy of terms without being awareof the need to browse different rows of the JPO matrix,and hence will not fall into the trap of failing to read thenext row of terms.

Other display options available with the STN thesauruscommand are +BT (Broader Term), +NT (NarrowerTerm) , +ALL and +TI (Title). The broader and narrowerterm commands will show only the terms above or belowthe selected term, respectively, whilst the title format dis-plays the portion of the hierarchy down to the selectedterm, with no narrower terms; this is effectively the firstsix rows of Table 6 in this instance, but if a theme codeor theme group is selected, the display would be shorter.

The STN thesaurus provides a flexible way of locatingthe definitions of F-terms obtained from a known record,but in the absence of a search engine on the actual textsof the terms, it is not ideal for developing a strategy directlyfrom an inventive concept. The ranking procedure is ofsome help with this, but depends upon the relevance of

FTCLA/FTERMFTERM CLASSIFICATION OF THE JAPANESE PATENT OFFICE4/FTERM. Chemistry4C/FTERM. . Medical Science4C076/FTERM. . . Medicinal preparations4C076/AA00/FTERM. . . . FORMS4C076/AA36/FTERM. . . . . Pills, tablets, lozenges, troches, or buccal agents4C076/AA37/FTERM. . . . . . Lozenges that act immediately4C076/AA38/FTERM. . . . . . Continuous or gradual-release preparations4C076/AA39/FTERM. . . . . . . Characterized by shape or structure4C076/AA40/FTERM. . . . . . . . Layered or laminated types4C076/AA41/FTERM. . . . . . . . Having discrete particles in a supporting matrix4C076/AA42/FTERM. . . . . . Coated preparations4C076/AA43/FTERM. . . . . . . Dragees4C076/AA44/FTERM. . . . . . . Coated tablets4C076/AA45/FTERM. . . . . . . . Enteric coated preparations4C076/AA46/FTERM. . . . . . Printed preparations4C076/AA47/FTERM. . . . . . Embossed, grooved, or perforated preparations4C076/AA48/FTERM. . . . . . Effervescent or foaming tablets4C076/AA49/FTERM. . . . . . Troches or buccal agents

S. Adams / World Patent Information 30 (2008) 5–20 17

the source hit set, and does not give much help to the userin determining the appropriate viewpoint(s) within a themecode. Unlike other data sets to which the ANALYZE com-mand is applied, the F-term system may be too granular toproduce a meaningful ranking to assist in selection of suit-able search terms, and such frequency analysis can cer-tainly never replace a detailed knowledge of the systemitself.

5.5. PatBase/Minesoft Ltd. + RWS

F-term data on PatBase are included for bibliographicrecords from mid-1984 to date, whilst the earliest date forrecord coverage is 1973. When F-terms are revised, correc-tion data are received for any records affected by thechanges (additions, modifications, deletions) and loadedinto the file. Since these revisions may also affectbibliographic records prior to 1984, a small number ofdocuments in the 1973–1983 range are now tagged withF-terms, but the majority are not.

The syntax for searching this field uses the field prefixJCT. The theme code and term code are separated by astroke symbol (/), just as for STN. The support files forPatBase indicate that the term code extensions have alsobeen loaded, and are separated from the main term codeby a further stroke. The full notation therefore becomes,for example, 2H091/FA14/X. Some of the definitions ofthese extensions have been loaded onto the system underthe definition for the viewpoint (i.e. the two-letter notation

Fig. 7. PatBase bibliographic rec

followed by 00), whilst others are available as a link fromPatBase to the PMGS. Searching an entire theme code ispossible without truncation, but obviously retrieves manythousands of hits and is not a preferred strategy in isola-tion. PatBase allows searching of the FI themes (i.e. F-termthemes with no associated term codes) but of course has nolink to corresponding term codes. There is no automaticfacility to retrieve hierarchically-related term codes, as onthe JPO site; the default in PatBase is to retrieve only thecode which has been input, whereas on the JPO site thisrequires the use of the $ prefix. Consequently, numbersof documents retrieved may differ between the two systemswhen apparently searching on the same code.

A ‘help’ link from the main search page yields a fewparagraphs of explanation of the F-term system, the searchsyntax and a link to the JPO for browsing. The ‘browseindex’ link connects directly to the home page of thePMGS (see Fig. 4). There is no separate search engineenabling a direct search of either theme code titles or termcode text, but the user could always apply method (c), con-ducting a sample search and analyzing the resulting recordsto determine relevance of the F-terms. In this context, thefamily record structure of PatBase is an advantage, sincea user could conduct a search in a more familiar classifica-tion system, then limit the results to only those cases with aJapanese family member, thus effectively creating an on-the-fly concordance.

Once a bibliographic record in PatBase containing F-terms has been located, each individual F-term is

ord with F-term highlighted.

18 S. Adams / World Patent Information 30 (2008) 5–20

hyperlinked to a listing of the relevant theme code – seeFig. 7. The term codes have been loaded in a listing format,clearly showing the dot level indents of successive terms.The full F-term which the user clicked appears at the top

Fig. 8. PatBase display of theme co

Fig. 9. PatBase them

of the page, which list the entire theme code in alphabeticorder of the viewpoints; the viewpoint of the selected termis highlighted to assist browsing of the term codes – seeFig. 8 which is the listing of theme code 5F041, and com-

de viewpoints and term codes.

e group display.

S. Adams / World Patent Information 30 (2008) 5–20 19

pare with the JPO matrix-style display in Fig. 3. It is pos-sible to navigate up to the theme group level, using the link‘Go Up One Level’ – see Fig. 9. A further ‘Browse classifi-cation’ link on both pages takes the user out of PatBase tothe JPO PMGS site, to view the theme code matrix.

As can be seen from Fig. 8, the PatBase display of theterm codes from a theme code is a vertical listing, similarto the STN display but retaining the JPO dot levels. Thisformat has both advantages and disadvantages. Theadvantage is that it removes the artificial 10-column formatof the JPO matrices. A vertical listing brings related terms(particularly narrower terms) to the attention of the user,which in the matrix display may be hidden on the nextrow and overlooked. A possible disadvantage, for userswho are familiar with other classification systems, is thatthe listing format may bias them towards selecting onlyone term from the whole theme, mentally applying the ‘lastpossible place’ concept from the IPC. This would be anincorrect approach, since the main strength of the F-termsystem is the ability to apply one or more terms from multi-

ple viewpoints within a theme, rather than being limited tothe one-dimensional classification approach.

Some examples were found in PatBase in which thebibliographic record included only a theme code, with nocorresponding term codes; for example, records for JP2004-173654-A2 and JP 2005-505747-T2 included referenceto theme codes in the 2G group (2G045, 2G055, 2G058).However, the facsimile front pages of these records showedonly F-terms from the 4B group (4B063, 4B024). Hence itseems likely that the electronic records are reflecting thecontinual process of revision of the F-term system, andwe are viewing older records which have so far been onlypartially re-indexed, by the addition of a new theme. Afew examples were found for which back-indexing with fullF-terms had been completed, whilst others had only beenpartially indexed with the new theme code.

6. Some further questions

Very little hard information exists in English concerningday-to-day operation of the F-term system, either in rela-tion to the policy for applying F-terms to documents, orthe revision of the system. The existence of a small numberof so-called ‘F-term explanations’ is of some help in under-standing how F-terms are selected for addition to a docu-ment record, but as noted above, these are available forless than 10% of the themes. Without a clear understandingof the policy for applying an F-term, the searcher maychoose inappropriate marks in their strategy. It wouldmarkedly improve the user’s confidence in the system ifEnglish translations could be produced of the ‘instructionsto classifiers’ or similar background documentation.

Of equal concern is the lack of information on the pro-gressive revision of F-terms. There is no published recordin English about whether a particular theme code, or indi-vidual term codes, have been phased out, nor where thatsubject matter is now indexed. The same problem faces

the user of the US patent classification; checking the frontpage for relevant class marks can lead to a highly mislead-ing strategy if a class has been re-arranged. In the case ofthe US system, there is at least the system of ClassificationOrders, which help the user to decide the current class incases where the issue class has been modified, but no anal-ogous documentation exists for F-terms. In the absence ofany convenient English-language means to view a currentelectronic record of the F-terms for a given record, search-ers may produce misleading strategies based on replaced ormodified term codes. The examples from PatBase illustratethe difficulty; the user is provided with no information as towhether the 2G045 terms should be regarded as supple-mentary indexing or replacing an earlier set of 4B groupterms.

In addition to the impact of the revision of F-terms onretrieval of current documents, it is likely that it will havean effect on retrieval from the back-file. Again, there isno regularly-published information in English concerningthe process of re-indexing when an F-term theme or termcode is replaced, although it is available for Japanesespeakers. Since F-terms have been applied back to thebeginning of the Japanese patent system, any revision ofan F-term theme code must affect many hundreds, if notthousands, of document records. At present, the user hasno information concerning the status of the re-indexingprocess for any given F-term, although the EPO [11] indi-cates that it can take up to two years before the process iscomplete. However, it is also understood that the JPOmaintains the ‘replaced’ F-term in use until the wholeback-file is processed. A complicating factor would be ifan F-term is merely revised, rather than closed; it thenmay become necessary to link the term with a suitable yearrange in order to produce an optimum retrieval strategy.This process of ‘time-slicing’ will be well-known to usersof the World Patent Index fragmentation code system,but at the time of writing there is no indication as towhether this approach is needed with F-terms. It would sig-nificantly enhance the usability of the system if detailednotices could be published by the JPO concerning whichtheme codes were being revised and the schedule for com-pletion of back-file re-indexing. These notices should alsobe made available to searchers by all the various commer-cial organisations which load F-term data.

This article has not directly addressed the question ofconstructing a search strategy, once suitable F-terms havebeen identified. This can be a complex process in itself, sim-ilar to other strategies using deep-indexing terms fromcommercial files such as Thomson’s Derwent World PatentIndex or the Uniterm� system from IFI/Claims. Some lit-erature on the process of F-terms strategy constructiondoes exist in English. Of particular note are the various pre-sentations from the European Patent Office’s periodicmeeting ‘‘East meets West’’. Unfortunately not all thematerial is yet available at the EPO website for these meet-ings. The author would commend in particular one work-shop presentation from the 2003 meeting for further

20 S. Adams / World Patent Information 30 (2008) 5–20

study [16], but at the time of writing this is not available onthe EPO website.

7. Summary

The availability of F-terms as a tool for non-Japanese-speaking searchers is a welcome addition to their armoury.The translation of the basic structure of the system into Eng-lish represents an enormous technical and intellectual effort.However, in order for Western searchers to become fullycompetent in their use of the system, some further back-ground information on the operation of F-terms is required.The help tools outlined in this paper are a good beginning,but the novice searcher should appreciate that there are vari-ations even between these tools. It is recommended thatstrategy generation should be approached with caution,using as many tools as possible to assist the process.

Acknowledgements

I am very grateful to Alan Engel (Paterra Inc.) and IreneSchellner (Japanese patent information section, EPO Vien-na) for helpful discussions during the preparation of thispaper, to Barry Dunne (SII) and the help-desk at theSTN UK Agency for advice on CAS policy and the STNthesaurus, and to Steve Lodge (RWS) and Ann Chapman(Minesoft) for advice on PatBase.

Parts of the subject matter from this paper were pre-sented at the PIUG Annual Conference, held in CostaMesa, California, 7–9th May 2007. The author acknowl-edges the permission of the conference organisers to ex-pand the material from the plenary session into thepresent article.

References

[1] Takei H. The Paperless Program in JPO. World Patent Inform

1989;11:14–24.

[2] Nakamura A. Paperless plan of the Japanese Patent Office and its

dissemination policy. World Patent Inform 1991;13:125–38.

[3] Schellner I. Japanese File Index classification and F-terms. World

Patent Inform 2002;24:197–201.

[4] Ueno S. JPO’s services and products concerning intellectual property

information dissemination. In: Proceedings of the PIUG Annual

Meeting 2002, Berkeley CA, USA, 20–24 April 2002.

[5] Nakata T. Far East meets West seminar, EPO Vienna, 25–26

November 2004. Supplementary material from Workshop C,

<http://www.european-patent-office.org/epidos/conf/jpinfo/2004/pro-

gramme/>.

[6] Kimura T. Views of users on contracting out searches In: Proceedings

of the AIPLA/FICPI Colloquium, ‘‘Patent Cooperation Treaty –

where we are and where we are going’’, 8–9 April 2003, Nice. <http://

www.aipla.org/Content/ContentGroups/Meetings_and_Events1/

International_Symposia 1/Kimura.pdf>.

[7] Alan Engel, Paterra Inc. – personal communication.

[8] <http://www.jpo.go.jp/shiryou/s_sonota/themecode.htm>.

[9] Anon. Paper presented to the annual meeting of the Standing

Advisory Committee before the EPO/Patent Documentation and

Information sub-committee (SACEPO/PDI). Vienna: EPO, March

2005.

[10] Anon. Patent Abstracts of Japan (PAJ) News, no. 40, 2006, p. 4–7.

[11] Irene Schellner, EPO Vienna – personal communication.

[12] <http://www5.ipdl.ncipi.go.jp/pmgs1/pmgs1/pmgs_E>, click on the

link ‘F-term’.

[13] <http://search.p4.patolis.co.jp/search_en.html>, click on ‘F-term’.

[14] <http://cxp.paterra.com/>, click on the link under the text ‘F-term

search system used by the Japan Patent Office’. Service withdrawn

January 2007.

[15] <http://patentinfo.european-patent-office.org/prod_serv/far_east/ja-

pan/index.en.php>, click on ‘Tips and Tricks for searching in

databases’, then select ‘Finding FI classes and F-terms for a Japanese

document’.

[16] Magrizo S; Alink M. Far East meets West seminar, EPO

Vienna, 23–24 October 2003. Supplementary material from

Workshop C.

Stephen Adams is founder and managing directorof Magister Ltd., an information and trainingconsultancy specialising in patents documenta-tion. He trained as a chemist at the University ofBristol, UK, followed by a Masters degree inInformation Science at City University, London.He has worked in technical information since1981, latterly with Zeneca Agrochemicals (nowSyngenta) as their principal patent searcher until1997. He has also been the editor of ‘‘Interna-tional Packaging Abstracts’’, a technical searcher

in the Ministry of Agriculture, Fisheries & Food in the UK, and Chair ofthe Patent and Trade Mark Group.