Standard ST.35

This Standard defines the formats to be used for the data exchange and processing of published patent information in mixed-mode (MM) form on ½ inch reel-to-reel and IBM 3480/90 cartridge magnetic tape (Mixed-Mode Magnetic Tape - MMMT). The Standard is based upon international standards and recommendations (ISO, WIPO, ITU-T (CCITT)); where necessary explicit references to these are made. The Standard provides for a device and layout independent presentation of patent documents with particular reference to exchange on magnetic tape.

The purpose of the Standard is, therefore, to provide a logical, independent, structure for patent document processing and for the data exchange of all data pertaining to one or more patents, be this text and/or image data. This means that this Standard may be used in place of WIPO standards: ST.30, “Recommendation concerning a standard magnetic tape format for the exchange in machine-readable form of bibliographic data, abstracts and full texts of patent documents” and ST.33, “Recommended standard format for data exchange of facsimile information of patent documents”.

Generally, the data exchanged may be used for building and updating bibliographic patent databases but it could also be used for full text and image patent databases - on any media: tape, hard disk, CD‑ROM, etc. In particular the Standard allows for the processing of the following data:

(a) Full text, or parts, of patent documents, including bibliographic data, recorded as character coded-data. It is strongly recommended that this data is marked up with SGML tags according to WIPO ST.32 (see below).

(b) Whole pages of documents represented as one image irrespective of their content (bibliographic data, text, or images).

(c) Data, within full-text documents, which cannot be recorded as character coded data, such as drawings, chemical formulae, complex tables, etc., may be treated as so-called embedded images (EMI’s) and are defined as frames.

Notes:

(i) Information about the text and image data must be given according to the identification and prefix information in “Appendix 2”;

(ii) For image data (2b,c above), the compression format recommended is ITU-T (CCITT) T.6: Group 4 (commonly known as ‘Fax Group 4’), other image formats are possible;

(iii) An alternative method of image data storage is given in “Appendix 4”: Tagged Image File Format (TIFF). Again, the compression format, within the TIFF record, recommended is ITU-T (CCITT) T.6: Group 4.

The following standards are of fundamental importance to this recommendation:

(a) WIPO Standard ST.32 - Recommendation for the markup of patent documents using SGML (Standard Generalized Markup Language).

(b) WIPO Standard ST.33 - Recommended standard format for data exchange of facsimile information of patent documents.

(Note: the prefix information in Appendix 2 is similar but not the same as ST.33, that used in ST.33 formed the basis of indexing for this Standard.)

(c) ISO 1001, Information processing - Magnetic tape labelling and file structure for information exchange.

(d) ISO 8879-1986, Information processing - Text and office systems - Standard Generalised Markup Language (SGML).

(e) ITU-T (CCITT), Blue Book, Volume VII – Fascicle VII.3 - Terminal Equipment and Protocols for Telematic Services - Recommendations T.0-T.63 – Recommendation T.6 - Facsimile Coding Schemes and Coding Control Functions for Group 4 Facsimile Apparatus (1984, amended 1988).

(f) Tagged Image File Format (TIFF). As administered by Microsoft Corp. on behalf of Aldus. (Detailed information about the structure of TIFF can be found in the Microsoft Windows Software Development Kit version 2.0 manual: ‘Windows Extensions’ chapters 1 through 7. It is available through the courtesy of Hewlett-Packard Company, Greely Division).

The appendices to this Recommendation contain the following:

• Appendix 1 – Character Set for Label and Record Prefix Coding (IBM EBCDIC)

• Appendix 2 – Prefix Item Definition

• Appendix 3 – Facsimile Coding Convention

• Appendix 4 – Tagged Image File Format (TIFF)

• Appendix 5 – Examples of Coding Mixed-Mode Patent Documents

For the purposes of this recommendation, the following specific definitions apply:

(a) RECORD LABEL: A short file on the magnetic tape containing the name and characteristics of the entire tape. A labelled tape starts with a volume label and each data file on the tape has a preceding header label and a following trailer label. In this respect this Standard follows ISO Standard 1001.

(b) PATENT DOCUMENT: The expression “patent document” comprises patents for invention, plant patents, inventors’ certificates, design patents, utility certificates, utility models, documents of addition thereto and published applications therefor. (A patent document may contain patent sub-documents - see below.)

(c) LOGICAL RECORD: A collection of all fields and data referring to one and the same patent document and being treated as one entity. A logical record may contain different document components.

(d) DOCUMENT COMPONENT: A collection of fields and data within a logical record (patent document) which may be treated as one entity, e.g., the full text of a patent document, one image.

(e) PATENT SUB-DOCUMENT: Within the variable part of a text component a patent document may contain sub-documents such as: bibliographic data, abstract, description, claims, drawings, search report.

(f) IMAGE COMPONENT: An image component will consist of one image: an embedded image, a drawing, a full page facsimile image (which may contain several images).

(g) BLOCK: A collection of records that can fit into 20,000 bytes.

(h) PHYSICAL RECORD: A collection of all fields and data within a document component which may be treated as one entity. When the size of a document component exceeds the block size of 20,000 bytes the data must be spanned over more than one physical record.

(i) SPANNING: a technique used for splitting a logical record into more than one physical record because the size of a logical record is bigger than the maximum size of a physical record.

This part of the Standard describes the recommended general logical layout and formatting of published patent documents exchanged on ½ inch reel-to-reel and IBM 3480/90 cartridge magnetic tape media described in Part 2, paragraph 14.

A logical record may not contain more than one patent document.

A dataset (file) with electronic patent documents may contain a sequence of logical records organised in, for example, ascending order of document identification.

Each logical record may contain information both in character-coded (text) and/or image form with respect to a patent document.

The following figure shows the general structure:

The Prefix Items may contain country code, document number, kind, etc., and any other general information which is useful for the processing of the variable part of the document. (See “Appendix 2”.)

Images can be stored in the variable part in various ways:

(a) Group 4 bitmap only;

(b) TIFF coded Group 4 bitmap.

The Standard allows for further extensions (CGM, JPEG, etc.) - in the prefix it is specified how the image is stored.

The sequence of the images may be identified by an eight byte sequence number. Usually this is composed of four characters for the page number followed by four characters for the frame number. Alternatively, it may be a consecutive number starting from one for each document (for example in cases where electronic publishing methods are used without reference to a paper document). The logical layout and relationship of these variable components can be illustrated as follows:

A typical patent document could be structured as follows (tagging according to WIPO ST.32):

In the example above, therefore, we have one patent document (logical record) containing three document components: the text data and two images. One component, the text data, contains six patent sub-documents. The link between the components is maintained by the <EMI> tags in the text data and the corresponding prefix information in the image data. The file name for the logical record, usually the patent application or publication number, can bind all components together. Further examples can be seen in “Appendix 5”.

The following section of the Standard shows how this data can be written onto magnetic tape (reel or cartridge) for data exchange. The use of other media, rather than magnetic tape, remains open for further study.

The standard specifications of the magnetic tape will be as follows:

(a) ½ inch, 9 track magnetic tape reel; or 18/36 track IBM 3480/90 tape cartridge;

(b) recording density of 6250 bpi for tape reel; 48 KB/inch for cartridge;

(c) standard labels for Volume, Header 1 and Header 2 as provided for in ISO Standard 1001, which, in respect to labelling, is compatible with IBM labels; user labels may also be added;

(d) label and record prefix encoding shall be in the Roman alphabet and in Arabic numerals, fully compatible with IBM EBCDIC and with the code set of characters shown in “Appendix 1”;

(e) one file can be spanned over multiple tape volumes in accordance with ISO 1001 standard.

ISO standard 1001, which defines magnetic tape labelling, label layout and usage, should be followed.

Each data set (file) may contain a collection of logical records each representing a patent document.

The following figure shows the general structure of a file:

The maximum number of physical records in a logical record is (2³² - 1). Therefore, the theoretical logical record size should be less than (2³² - 1) x 19,996 positions (the maximum block size 20,000 minus the block length indicator of 4 bytes).

The total maximum physical record size is 19,996 including record length indicator of 4 bytes.

A physical record has the following specification:

(a) the recording mode is variable blocked; that is: a block may contain several physical records;

(b) the maximum block size is 20,000, including the block length indicator of 4 bytes;

(c) the total maximum physical record size is 19,996 including record length indicator of 4 bytes;

(d) the physical record layout may be schematically shown as:

Each physical record starts with a prefix of 256 bytes. This includes the record descriptor word (binary record length indicator of four bytes), however, depending on the system environment, this field may not be directly accessible by some application programs, e.g., IBM OS/VS COBOL. The prefix is defined in “Appendix 2”.

For each component there is created at least one physical record. When the size of a component exceeds the maximum physical length then a spanning technique, described below, is used to create more physical records.

Physical records are created on magnetic tape using the variable record organisation.

A spanning technique is necessary for components due to the inability of magnetic tape devices to handle, under normal circumstances, physical records in excess of 20,000 characters. The record prefix contains two items concerning the spanning technique, viz. item 9 and 19 (component record sequence number and highest document component record sequence number, both two byte fields, see “Appendix 2”, to be used by application programs).

Examples of coding mixed-mode patent documents are given in “Appendix 5”.

ASCII Character Set Usage

The usage of the ASCII character set, as an alternative to the EBCDIC character set, for prefix items is allowed and does not affect field sizes. In other words the only requirement is an indicator saying that the data is in ASCII and not in EBCDIC.

It is recommended to use a one byte field and store an “A” in ASCII for ASCII coded data or an “E” in EBCDIC for EBCDIC coded data. In the prefix information this can be position 19 (i.e., the first byte of item 6: “Others”)

For the fields coded as binary data it is recommended to have this data available in character format, because of the differences in interpretation of binary data in various operating systems (e.g., UNIX). Here it is also proposed to use “free” fields from items “Others”. The fields coded in binary are:

(a) Item 9: Document component sequence number, 2 bytes. The maximum value in practice shall never exceed 10,000 (-1), so a field length of 4 characters (in ASCII or EBCDIC) will be sufficient. The space can be found in item 23: position 107-110.

(b) Item 18: Highest document record sequence number, 4 bytes. The maximum value shall never exceed 1,000,000 (‑1), so a field of 6 characters (in ASCII or EBCDIC) will be sufficient. The space can be found in item 23 (others exchange use): position 111‑116.

(c) Item 19: Highest document component sequence number, 2 bytes. The maximum value in practice shall never exceed 10,000 (-1), so a field length of 4 characters (in ASCII or EBCDIC) will be sufficient. The space can be found in item 23: position 117‑120.

(d) Item 49: Length of variable data field, 2 bytes. Although the value can be derived from item 1 (record length in character format - prefix length), it can also be stored in item 6 with the same length of item 1, length 5 characters: position 20‑24.

Notes:

In practice, if data is delivered on magnetic tape, the tape label information would normally be in EBCDIC (ISO 1001), even if the prefix information is in ASCII. In such cases the tape label information may be ignored or, more usually, automatically processed by receiving systems.

It is recommend that if the prefix information is in ASCII then the variable text data (SGML data) is also in ASCII and vice versa for EBCDIC data.

It is recommended that, before information is regularly exchanged between Offices, full discussion should take place as to the exact manner in which this Recommended Standard is implemented, particularly in connection with the information to be recorded under “Other” item numbers 6, 11, 12, 23, 24, 34, 35, 47, 48 of the prefix definition given in “Appendix 2”.