Appendix_I_-_Image_Specs.doc

National Library of Medicine

Article Level Historical Journal Scanning and XML Preparation for PubMed Central Appendix I

Image Specifications and Functional Requirements for Citation Capture Version 1.0 April 2015 National Library of Medicine National Institutes of Health/Health & Human Services

8600 Rockville Pike

Bethesda MD 20894

Purpose and Scope of this Document The purpose of this publication is to document all requirements for image capture, article identification, and citation creation for the materials scanned as part of the PubMed Central Back Issue Scanning Project. It is subject to change, and may specify requirements that are only pertinent to records created for NLM’s PubMed Central database.

Table of Contents

41.0 Project Overview

2.0 Project Workflow 3.0 Quality Assurance 4.0 Source Description 5.0 Target Database: General Specifications 6.0 Target Database: Image Specifications 7.0 Target Database: XML Element Description 8.0 Appendix A: Journal Style Guides 9.0 Appendix B: Accepted Rework (AR) Procedure

1.0 Project Overview

Summary

This project involves capturing bibliographic citation data (XML) and full article text (OCR) as well as creating article level PDF files and full page images of each page (TIFF) that will be added to the customer database containing a digital archive of life sciences journal literature.

The customer database already contains bibliographic citation data for some of the articles. The customer will provide a list of articles that are already in their database and for which the XML data does not need to be created.

Definitions

Article-Level

The term article-level defines a set of pages of the same page type (e.g. article, administrative content, advertisement and table of content pages) and grouped according to the logical boundaries of the item in the source. For example, all pages of a specific article are grouped together into a single article-level item. Similarly, all pages of administrative content are grouped together into a single article-level item.

1.1.1 Exclusion List

A customer-suppled XML file identifying articles that already have metadata in PubMed.

1.1.2 PMID

A unique number assigned to each PubMed citation.

1.1.3 Schematron

Schematron (http://www.schematron.com/) is a language used to enforce structures in XML documents more specific than are defined in a validation schema, such as a DTD. For this project, the Schematron provided by the customer enforces a strict subset of the rules defined in the JATS Archiving and Interchange v1.0 DTD.

1.1.4 Section 508 and Section 508-Compliance

Section 508 of the Rehabilitation Act of 1973 (29 U.S.C. 794d) addresses the requirement of Federal agencies to meet specific accessibility standards regarding electronic and information technologies for persons with disabilities. Within the scope of this project, compliance is required for the PDF deliverables and documentation for the customer. Compliance is addressed in Section 6.11 and is limited to the requirements outlined in the following clauses of the Government Product/Service Accessibility Template (GPAT) accompanying this project’s solicitation documents:

· 1194.22a: PDF deliverables will be tagged and include alternative text for images as described in 6.11.1.

· 1194.31a and 1194.31b: PDF deliverables will include embedded OCR text to enable the PDFs to be read out loud by assistive technology. Specific requirements are described in 6.11.1

1194.41a and 1194.41c: Project documentation for the customer will be made available in an alternate, accessible form, should it be requested.

Project Deliverables

The deliverables for this project include:

A 400-ppi 24-bit color TIFF for each issue page. Issue pages that contain text from more than one article (see Item 5.2.1) are duplicated in the corresponding directories for each article.

A cropped color TIFF image of each article illustration at 400-ppi, 24-bit color (LZW compression). Illustration images are delivered only for article page types (see Item 5.2.1).

A Section 508-compliant article-level PDF file of one or more related pages with composite images for pages that contain a mixture of 8-bit grayscale page text and grayscale or color illustrations. Pages in the PDF appear in reading order sequence, typically the same sequence as the original source. This type of PDF file is delivered only for article page types (see Item 5.2.1).

For full-color articles, the article-level PDF will contain color page images.

For TOCs appearing on full-color front covers, the TOC-level PDF will contain color page images, when requested by the customer-provided Journal Style Guides OR by the contractor specifically.

A Section 508-compliant article-level PDF file of one or more related pages comprised of the grayscale full-page images. Pages in the PDF appear in reading order sequence, typically the same sequence as the original source. This type of PDF is delivered only for non-article page types (see Items 5.2.1.3, 5.2.4 and 5.2.5)

An article-level ASCII file of the text of each article page, created using OCR technology (unedited). If any page in the article contains text from more than one article, the OCR text file is edited to exclude any text belonging to other articles. Autozoning generally achieves column recognition. However, in batches that this is not consistently achieved, the OCR files should be corrected to obtain basic page column recognition.

An article-level XML-tagged document for each article for which the customer does not already have citation data

A “place-holder” XML file for each article for which the customer already has citation data

A manifest file for each package of delivered files listing the name and MD5 checksum of each file

A file listing the PMID of any articles contained in the customer-supplied Exclusion List, but not present in the deliverable (see Item 1.6)

An inventory file mapping the contractor-assigned page image filenames (see Item 5.6.2) to source page number.

The inventory file will be issue-specific and contain a header with issue-level information. The format of the header is as follows:

<path="ppp"><journal="jjj"><vol="vvv"><issue="iii"><date="ddd"> where ppp:

path to the top level directory for the journal issue jjj:

journal code (see Item 5.6.1.1) vvv:

volume number as per source iii:

issue number as per source ddd:

the issue date as per source

Source Material

1.1.5 The source material consists of paper journals, bound and unbound. The expected volume is is between 777,000 to 1.5 million pages delivered in phases.

1.1.6 NLM source material provided for digitization shall be kept in climate-controlled processing and storage areas with temperature maintained in the range of 72°F (+/- 2°F) and relative humidity in the range of 50% (+/- 5%) without uncontrolled fluctuation to extremes outside of these ranges. Storage conditions shall be similarly maintained in climate-controlled spaces of vehicles used to transport NLM material from the Library and back.

Project Workflow: Overview

Customer prepares source materials for pick up.

The shipments will be picked up by the contractor from the customer site and brought to the contractor.

The contractor confirms content against customer-supplied packing lists.

The contractor will review each page of the source material and report any source discrepancies.

The contractor will divide source material into batches and provide inventory and schedule information to customer within five (5) business days after receipt.

The contractor batches data and assigns batch number to all issues from the customer-supplied packing lists.

Article Selection

The majority of the articles will require contractor generated XML. The customer will supply a list of articles, referred to as the Exclusion List, that already have metadata in PubMed. The contractor will not be required to generate XML for those articles, with the exception of adding abstracts, if there are any associated with an article. This list will be reconciled against the article inventory created during production. Articles excluded from XML creation are identified as satisfying one of the following parameters:

If <Item Name="HasAbstract" Type="Integer"> is '0' and the corresponding articles does NOT have an abstract in the source

If <Item Name="HasAbstract" Type="Integer"> is '1' and the corresponding articles DO have an abstract in the source See example of both cases, below:

Any article either not satisfying the above parameter or not listed in the customer-supplied file will be converted.

NOTE: Articles are excluded ONLY on the basis of the parameters described above. Any differences in the number of authors presented in either the article or the Exclusion List does not have any bearing on the exclusion of articles.

For excluded articles, an XML file will be generated containing only the customer-assigned <Id> number within an Article Identifier element (see Item 7.5.2)

These guidelines are used for resolving discrepancies between article boundaries as defined in the source or Journal Style Guide requirements and as defined by the Exclusion List:

General rule:

Follow the TOC for article boundaries/definition and the rules below. The specific Journal Style Guide if it lists an exception to the TOC rule will always override the rules below.

Case 1:

TOC requires multiple articles AND the Exclusion List contains single article encompassing all pages of multiple articles.

Create the individual articles per the TOC or Journal Style Guide. Match the first individual article to the Exclusion List item encompassing all pages.

Special Case:

In some cases, relying on pages cited in the exclusion item alone is not enough to determine if the exclusion item contains one or multiple articles. This is especially common in cases of letters followed by replies.

These cases can only be resolved by referring to the letter title and the authors if available. When letters and following reply letters are grouped together as a single item in the Exclusion List, match the exclusion item with the first letter. All following reply letters are not linked to the Exclusion List item and therefore are not excluded.

Case 2:

TOC requires one article AND the Exclusion List contains multiple entries for individual articles within the section.

Split the section into individual articles. Match the article as cited in the entry from the Exclusion List. This will result in multiple articles being created for each item in the section, where some or all of the articles may be matched to the Exclusion List.

Example:

TOC:

100-150

Exclusion List:

article 1: 101-120 article 2: 140-150

XML:

article 1: 100 (individual article XML) article 2: 101-120 (<article-id> only; matched to Exclusion List) article 3: 121-139 (individual article XML) article 4 : 140-150 (<article-id> only; matched to Exclusion List)

In this example, a single article is listed in the TOC as 100-150. However, the Exclusion List contains entries for two of the cited articles (101-120 and 140-150). Therefore, these two articles need to be matched accordingly. Additionally, articles contained in the section but NOT cited in the Exclusion List should still be separated out as individual articles (100 and 121-139).

Source Discrepancies

NLM has performed a condition review on all materials included in this project. Much of the material is old, and some of it is in poor condition which may require special care in scanning to prevent further damage. Unless the volume cannot be scanned at all, the project will proceed with the materials as delivered.

During the inventory checking process for each batch, the contractor will identify any source discrepancies caused by missing volumes or issues that are not indicated on the customer’s style guides and/or packing lists.

Contractor will create targets only for missing pages within an issue.

The contractor will inform the Customer of all source discrepancies within each title.

Customer Delivery Requirements

All files are to be delivered by FTP. The source documents should be held in safe and environmentally appropriate custody (see Sec. 1.4, above) and need to be returned to the customer when the files are delivered.

Files are to be in a consistent naming structure as determined by the customer. A separate directory will be delivered for each article that contains all the associated files for that article.

For each batch of data, the following deliverables will be sent:

· Deliverables listed in Item 1.3

· Batch delivery log identifying any article constructs that were not specifically defined in the Journal Style Guide with an explanation of their treatment in the delivered files

2.0 Project Workflow

Shipment Preparation by Customer

Customer will review the source material for completeness and usability for scanning. The source materials will then be collated, boxed and labeled for shipment.

Each shipment will include a packing list (see appendix XX) of the contents which will contain the following:

· Journal Title

· Volume/Part/Year Information

· Total Number of Volumes/Issues

· Estimated Total Number of Pages for each Journal

The customer will also e-mail an electronic packing list to the contractor, containing the same information as above, which the contractor will use to inform the customer of the batch number that the particular journal shipment has been assigned to. (See below, 2.3.2).

Shipment Preparation and Inventory by the contractor The contractor will retrieve the shipment from the customer.

The contractor will verify the shipment contents against the customer-supplied packing list and ensure that every issue is present.

Perform a full inventory of all source material in the shipment, including actual page counts.

Identify any missing volumes or issues not included in the customer-supplied packing list as source discrepancies

Provide inventory details and source discrepancies to the customer.

Batch Inventory and Schedule

The contractor will use the full inventory information provided by the customer to divide the content of the shipment into production batches of approximately 3,000 pages. The contractor will assign a batch number to each batch.

Provide batch inventory and production schedule to customer.

The batch inventory will detail the contents of each batch including:

· Contractor assigned batch number

· Journal title of batch content

· Volume/issue numbers of batch content

· Issue-level ID number supplied by Customer in source packing list

The production schedule will detail the schedule for the following:

· Delivery dates for each batch (referenced by the contractor assigned batch number)

· Adjusted delivery dates, if needed

Source Material Problems The contractor will report any missing volumes or issues not reported by the customer. Discrepancies should not be reported for damaged pages of articles and administrative material, missing covers and TOCs not noted on the packing list, and missing pages determined in the full inventory conducted by the contractor. No missing advertising pages should be reported as a discrepancy. .

Production Inventory

The contractor will perform a complete inventory of the shipment and compare it to the customer supplied inventory.

3.0 Quality Assurance

Overview

The contractor will employ methods such as random sampling, or biased sampling focused on areas that may be problematic and internal testing before release to customer in order to maintain the following specific quality measures for completed and delivered work product.

Product QA Acceptance Criteria

Criteria (per sample)

· Incorrect application of XML element

· Includes incorrect article type assigned as defined in specifications and procedures outlined in Section 7.2.1.1 99.95%

· XML not valid per DTD

· XML not valid per Schematron 100%

· Image skewed on visual inspection

· Image not consistent with source (e.g., inconsistent boundaries, pixilation, distortion, lack of sharpness and detail, etc) 99%

5A

Inventory error

· Missing or duplicate file (XML, OCR, image, etc)

· Citation not created for non-excluded item

· Images not sequenced in “reading order”

· PDF exceeds 60-page limit

· PDF is not Section 508-compliant

· OCR text does not correspond to article boundary

· Individual figure delivered as multiple images 100%

· Article boundaries do not conform to specifications and procedures outlined in this document, Exclusion List or Style Guide, whichever is applicable

· Incorrect article type assigned as per specifications and procedures outlined in Section 5.2 – i.e. page types 99.5%

For rejected batches, all errors that failed the batch will be corrected and the batch will be reviewed for additional instances of those error types.

The customer may request that a batch be reviewed and corrected if either or both of the following conditions are present:

a. The error is clearly in conflict with the specifications and procedures outlined in this document, and

b. The error is a systematic problem that occurs throughout the batch.

Even though the customer did not identify enough errors to fail such batches, the batch will still be considered rejected and will require correction and re-delivery.

Product QA Methodology

Product QA sampling is based on the primary deliverable for the project: the article PDF file. The article PDF file is the primary deliverable as it is most representative of all deliverables for this project (XML data, full-page images, illustration images and OCR text).

A randomly generated sample of 7% of the total number of article PDF files in a batch will undergo Product QA by the customer. All acceptance criteria are determined based on the contents of this sample. Results that are below the stated accuracy requirement for any acceptance criteria will cause the batch to be rejected.

XML File Acceptance Determination

Character accuracy

Text contained in the article XML files in the sample is proofed against the original source material. The total number of incorrect or missing characters is measured against the total byte count in the sample, the result is multiplied against “100” to obtain the accuracy rate.:

(Total Byte Count) – (No. of Incorrect Characters)

Total Byte Count Tag content accuracy

Elements in the article XML files in the sample are checked for correct content. An example of an error would be a contributor surname (<surname>) is incorrectly “tagged” as contributor given name (<given-names>). The total number of elements with incorrect or missing content is measured against the total number of elements in the sample, the result is multiplied against “100” to obtain the accuracy rate.

(Total No. of Elements) – (No. of Incorrect Elements)

Total No. of Elements Structural accuracy

All XML files in the sample are parsed and validated against the NISO JATS Journal Archiving and Interchange v1.0 DTD as well as the project Schematron.

Inventory error

An article XML file is required to be generated for each article in the sample:

· Articles identified as excluded still require the minimally tagged XML file as per project specifications.

· Articles not identified as excluded must have a fully tagged XML file.

Full-page Image Acceptance Determination

Unsatisfactory Quality

Full-page images in the sample are reviewed for the following:

· Page image is skewed upon visual inspection

· Page image boundaries not consistent with source

· Image contains visible pixilation, distortion, color distortion or lack of sharpness and detail

The total number of poor quality page images is measured against the total number of page images in the sample, the result is multiplied against “100” to obtain the accuracy rate:

(Total No. of Page Images) – (No. of Poor Quality Images)

Total No. of Page Images Inventory error

A page image is required to be generated for each page in the sample:

· Page images must appear in same sequence as source

· Page images cannot be duplicate or missing

Illustration Image Acceptance Determination

Unsatisfactory Quality

Illustration images in the sample are reviewed for the following:

· Illustration image boundaries not consistent with source

· Image contains visible pixilation, distortion, color distortion or lack of sharpness and detail

The total number of poor quality illustration images is measured against the total number of illustration images in the sample, the result is multiplied against “100” to obtain the accuracy rate.

(Total No. of Illus. Images) – (No. of Poor Quality Images)

Total No. of Illus. Images Inventory error

An illustration image is required to be generated for each illustration in the sample.

· Multipart illustration images must be created as per specifications

PDF file Acceptance Determination

Inventory error

An article PDF is required to be generated for each article in the sample.

· Content of article PDF must correspond to article boundary as per source

· PDF pages must appear in correct sequence as per specifications (see Item 6.11.4)

· PDF page contains OCR text

· PDF must not exceed size restriction as per specifications

· PDF must contain accessibility tagging and be Section 508-compliant (see Item 6.11.1) OCR file Acceptance Determination

Unsatisfactory quality

OCR files in the sample must have been generated using auto-zoning.

Inventory Error

An OCR file is required to be generated for each article in the sample. The OCR text must correspond to article boundary

General Acceptance Determination

All files must be assigned the correct page type (see Item 5.1)

4.0 Source Description

General

The source material for this project consists of bound volumes and some single issues of various medical journals. A designated portion of the journal run will be converted in this project. The material can span from as far in the past as the mid-1700s.

Source Quality

Variation

Source quality varies greatly. Much of the material is quite old, and some is very brittle.. All of these variations must be accounted for throughout the entire production process.

· Older source material can be very brittle and is yellowish in appearance.

· More recent source material is usually of better quality, but will also be very glossy in appearance.

· Bound copies of multiple issues may be printed on extremely thin paper. Special care must be taken to avoid damaging the source material, particularly during the imaging process.

Note: Scanning should proceed unless the contractor finds that a volume cannot be scanned due to its physical condition. The contractor shall contact the COR in these instances for further instruction.

General Content

Regardless of the binding format, most issues will be relatively consistent in content. In addition to articles (which are the majority of an issue), an issue will often contain the following:

· Covers

· Table of Contents

· Articles

· Administrative Material (author index, article index, keyword index, editorial mastheads)

· Advertisements (including advertising index) Issue Covers

The covers of most of the issues (or the first page of an issue in bound copies) contain issue identification information:

The date information is typically located on the issue cover (see Item 4.4.4). However, if a cover is not available, follow the instructions on the Journal Style Guide.

Journal title

The journal title is very prominently featured on the issue cover.

Volume number

The volume number can be identified with a keyword such as "Volume" or "Vol." The number may appear in either Roman or Arabic numerals. While, other issue identifiers may be optional, the volume number is expected to always be present.

Issue number

The issue number can be identified with a keyword such as "Issue" or "Iss." The number may appear in either Roman or Arabic numerals. The issue number is not always present in the source.

Issue date

The issue date can appear in various formats, containing any of the following information:

· Day information (often in numeric form only)

· Month information (often spelled-out or abbreviated)

· Year information (often in 4-digit numeric form)

· Season information (often present instead of month information)

The sequence in which this date information appears is varied. However, it is expected that the most common format would be "Month Day, Year", i.e. "November 7, 1926".

Table of Content Pages

The table of content pages lists the content of an individual issue

Issue Information

These pages can contain issue identification information similar to that often found in the issue covers.

Article Information

A single article entry can contain article title, article authors, first article page and/or a brief description. Additionally, articles can be grouped into subject categories; each category would have a distinct heading or title.

Multiple Table of Contents Pages

Source may be received as either individual issues or bound volumes (containing multiple issues from the same volume). Within these two types of source there may be multiple table of contents pages. These types of table of contents pages are defined as:

· Cumulative TOC: Relevant to the contents of an entire volume

· Individual TOC: Relevant to the contents of a single issue

If there are individual TOCs within a bound volume, then the individual TOC should be made part of the corresponding issue. The individual TOC would be placed in the TOC category.

If there is a cumulative TOC within a bound volume, then the cumulative TOC should be made part of the issue in which it was published or bound. Do NOT capture the cumulative TOC for each issue within the bound volume. The cumulative TOC should be captured as the issue TOC if there is no individual issue TOC. If there are multiple types of TOCs present in an issue, treat the individual TOC as the issue TOC and place the cumulative TOC with the administrative material.

Front Matter Pages

Issue Information

These pages can contain issue identification information similar to that often found in the issue covers.

Publisher information

The copyright often contains publisher information, including publisher name and location.

Article Pages

The first page of an article contains the majority of the bibliographic metadata, including:

· Article title

· Article authors

· Article footnotes

· Article copyright

· Article abstract

Errata/Correction/Retraction Articles

These articles are short paragraphs indicating corrections made for previously published articles. They must contain volume and page references to the previously published article. Each errata item is treated as a separate article.

Multi-part/Continuation Articles

These articles are long articles of any type that are published in shorter parts across multiple issues of a publication. They are indicated by text appearing at the beginning and/or the end of the article, using phrases like “to be continued”, “continues”, or “continued from”. The later parts of the article usually carry the citation information of the previous part. Each part is treated as a separate article.

5.0 Target Database: General Specifications

Specifications Hierarchy

The specifications and procedures outlined in this documentprovide the general specifications to be followed for this project. However, the customer also provides Journal Style Guides which detail title-specific requirements and an Exclusion List (used for determining article exclusion) as supplementary material. When there is a conflict in the requirements between these three types of documents, then the following hierarchy should be used to resolve the conflict:

· Journal Style Guide

· Exclusion List

· Specifications and procedures outlined in this document Issue Pages and Page Types

All issue pages are scanned as full page images (see Item 1.3.1), including outside/inside covers, tables of contents, articles and advertisements. Each page is categorized as containing any of the page types listed below.

Page types determine how a page is converted and delivered. Pages containing multiple page types are duplicated for each of the corresponding page types. For example, if a page contains Table of Contents, Advertising and Article text, the page is identified as having “article” page type, “advertising” page type and “table of contents” page type. This page will appear in the deliverable for the article, the table of contents and advertising.

There is no exception to this rule, all page are duplicated for each page type.

Articles

Articles comprise the majority of an issue. The term is used to categorize all full-length material, as well as shorter articles such as meeting reports, commentaries, reviews, etc. With few exceptions, most entries listed in the issue table of contents are considered to be articles.

Article pages are defined as pages that contain ANY article text.

Grouped Articles

· Some articles are actually groups of separate items. In these instances, each individual item is considered an individual article.

Grouped articles can often be identified by the presence of discrete sections of text with specific set of contributor(s). For example, a “Book Reviews” section may contain multiple reviews, each written by a different contributor. In these cases, each review would be a separate article.

· Each individual letter in a “Letters to the Editor” section would be treated as a single article. (Exhibit 1) Additionally, any brief editorial comment or author comment following a letter is treated as part of the preceding letter and not as a separate article. (Exhibit 2)

· Each correction item in an “Errata” section would be treated as a single article. (Exhibit 3)

· Each individual book review in a “Book Reviews” section would be treated as a single article. (Exhibit 4)

· Each obituary is a detailed write up on an individual’s accomplishments and has a contributor so would be a separate article. (Exhibit 5)

· In some cases, the customer supplied Journal Style Guide may require that grouped articles should be treated as a single article, for example, a book review section containing multiple books reviews could be treated as a single article on an exception bases for a particular title.

Poster Abstracts

Some journal issues will contain poster abstracts of a conference proceeding. These pages contain images reproduced from a poster at the conference or text capturing summaries of the posters. These are brief abstracts summarizing presentations made during the conference proceedings. Several poster abstracts may appear grouped together on one or more pages.

The grouped abstracts may not be typeset text as seen for regular article text. Instead, the poster abstracts may be reproductions of original documents which are reduced in size in order to fit multiple poster abstracts on a single page.

Poster abstracts are not treated as grouped articles. They are not split into individual articles. A group of poster abstracts is usually listed as a single entry in the issue table of contents. This entry is used to determine article boundaries and titles. (Exhibit 6) Proceedings Articles

Some journal issues will contain research articles presented at a conference proceeding. This is a series of articles that cover a single conference proceeding. This series include proceedings abstracts, these differ from poster abstracts, as the text is usually the same as the typeset text for a regular article.

Proceedings articles are generally, but not always treated as part of a single article. The entire section detailing the proceedings, including the abstracts, schedules, timetable, President’s Address, etc. are all grouped together as a single article.

Sometimes a PMID occurs in the Exclusion List which applies to the entire proceedings section of an issue, or to a subset of the pages included in the proceedings section. Use the PMID in the <article-id> tag in the following cases:

· For the single article where no division of the proceedings

· For the first article where the proceedings is divided into sections

Proceedings can appear as either grouped into a single article or split into sections. The customer-provided Journal Style Guide will often provide details on the treatment of proceedings articles.

In general, if the proceedings articles are listed in the TOC, then treat the entire set of proceedings articles as a single article. However, if the proceedings articles comprise the entirety of an issue, then divided the proceedings article into appropriate sections (e.g. by date, subject divisions, or association sections) (Exhibits 7-8)

Follow-up Articles

These are articles that are related to the preceding article. These articles are often titled as “Replies” or “Discussions”. These follow-up articles are treated as individual articles, although they will be linked to the preceding article via the <related-article> element (see Item 7.11) (Exhibit 9) Obituaries and Death Notices

A distinction must be made between obituaries and death notices. Obituaries are captured as individual articles, whereas death notices are captured as part of administrative material.

An obituary is typically a lengthy article detailing an individual’s accomplishments. An obituary article would typically have a contributor.

However, death notices are lists of individuals who have died. The list usually consists of names, birth and death dates and place of residency. Death notices would NOT typically have contributors.

(Exhibits 5, 10) Table of Contents

These pages contain the table of contents page(s) of the journal issue. The tables of contents can be issue-level, volume-level, subject specific, or cumulative for several volumes.

If an issue has both a cumulative and individual TOC (see Item 4.5.3), then the individual TOC is treated as “Table of Contents” and the cumulative TOC is treated as Administrative Content (Exhibit 11) If an issue has only a cumulative TOC, then the cumulative TOC is treated as “Table of Contents”.

If an issue has only an individual TOC, then the individual TOC is treated as “Table of Contents”.

Advertising indexes and TOCs of forthcoming issues are not treated as “Table of Contents”. Advertising indexes are treated as “Advertisement” (see Item 5.2.4) and TOCs of forthcoming issues are treated as “Administrative” (see Item 5.2.3) (Exhibit 12) Administrative Content

These pages contain various types of content, including

· Journal Masthead

· Editorial Board

· Subject or Author Indexes (not including Advertising Indexes)

· Instructions to Authors

· Administrative Announcements for Society Members

· Meeting Notices (different from meeting reports)

· Death Notices (lists of individuals who have died; usually consisting of names, birth and death dates, and place of residency)

· Calls for Papers

· Course Announcements

· List of Books (e.g. books received, selected reading, selected titles etc.)

· Book Notices (usually for review or consideration).

· Annual Report

· Title Pages

Administrative Content and the TOC

Some administrative content, such as announcements of upcoming conferences, new books, etc., can be hard to distinguish from advertisements.

All announcement material (including book announcements, meeting announcements, course announcements and drug product announcements) should be placed with advertisements unless listed on the TOC in which case it can be placed with administrative material.

Book notices/new book lists that appear on the TOC will go with administrative material. If not on the TOC, it should be placed with advertising.

Indexes Indexes are usually converted as Administrative Content. However, there is an article type for “index” (see Item 7.2.1.1). An index is treated as an article only when the issue is a supplement that comprises only indexes, i.e., the issue has no "regular" article content.

A cumulative index for a volume should be placed in the admin section of the issue with which it is bound (if the cumulative appears with the first issue of the volume, it should be placed in the admin section of the first issue; likewise, if the cumulative index appears with the last issue of the volume, it should be placed in the admin section of the last issue).

If there is a cumulative index covering multiple volumes, this should be treated as its own issue. If there are cumulative indexes containing multiple volumes associated with one issue, then this would be a mistake on the packing list and should be raised as a source discrepancy.

(Exhibit 13)

Subscriptions

All subscription material should be treated as Advertisement.

Advertisement

These pages are all the page advertisements in the issue, including partial and full-page ads. “Advertiser Index” or “Index to Advertisements” should also be captured as Advertisement (Exhibit 12).

Issue Covers

These pages are the outside and inside covers of the journal issue.

Blank Covers

Blank covers are not deliverable.

Outside Front Cover

Outside front covers containing material of other page types, such as table of contents, are duplicated in the appropriate file type, e.g. cover and TOC. (Exhibit 14) Source Page Numbers

The source page number for each page must be recorded in the document inventory. In some cases, the number may not be explicitly present on the source page. Use the following guidelines to resolve such cases.

If the source page number is inferable, then the inferred page number must be created within square brackets "[ ]". Inference of page numbers can be based on either the surrounding page numbers or on the page number provided in the table of contents entry. The table of contents entry is useful for inferring page numbers, when the missing source page number is the first page of the article.

Additionally, if the missing source page number occurs on the first article page of an issue and the last numbered page of the previous issue is known, then it is acceptable to infer the missing source page number using the sequence of the last known page number of the previous issue. For example, if the first page of the first article in an issue is unnumbered, but the last numbered page in the previous issue is 157, then you may infer the missing page number as [158] (see Item 5.3.2 for additional clarification).

When inferring source page numbers it is important to preserve the sequence of numbering within a single article-level item (see Item 1.2.1) or series of pages of the same the page type (see Item 5.1). In some cases, it will not be possible to infer the page number. These cases are discussed in later paragraphs.

For example, source page numbers appear as follows:

1st Administrative Content Page:

2nd Administrative Content Page:

3rd Administrative Content Page:

4th Administrative Content Page:

[unnumbered] 1st Article Page:

[unnumbered] 2nd Article Page [unnumbered] 3rd Article Page 4th Article Page:

[unnumbered] 5th Article Page:

In this example, there are four unnumbered pages. The unnumbered administrative content page would be numbered [4] since it follows in the sequence of administrative content page numbers. The next two pages are article pages that precede a page whose number follows in a sequence from a previous issue. Their pages cannot be inferred since neither [5] and [6] nor [156] and [157] would be correct. (The former would imply that these are part of the administrative section, and the latter may conflict with pages from the previous issue.) The fourth article page would be numbered [159].

If the source page number CANNOT be inferred but is preceded by a numbered page within the same page type, then the last known source page number is keyed within square brackets and a sequential number suffix:

[ssss]-a where:

ssss:

is the last source page number (inferable or explicit) a:

is the sequential number of the page following

EXAMPLE: There are two pages without a source page number between Pages 234 and 235. These pages would be numbered, respectively as "234-1" and "234-2".

EXAMPLE: There is one page without a source page number between Pages [65] and 66 (where Page 65 is an inferred page). The unnumbered page would be keyed as "[65]-1".

If the source page number CANNOT be inferred and the unknown page number precedes the first page in a sequence of pages whose numbering may have been continued from a previous issue, use the page number of the first numbered page in this page type in square brackets followed by bn where “b” is the literal “b” character and n is a sequential number. In the following example:

1st Administrative Content Page:

2nd Administrative Content Page:

3rd Administrative Content Page:

1st Article Page:

[unnumbered] 2nd Article Page [unnumbered] 3rd Article Page

The two article pages preceding 158 have unnumbered pages that cannot be inferred because the number 158 is continued from a previous issue. The two unnumbered pages should be respectively numbered [158]-b1 and [158]-b2.

Note that in the case, we are assuming the last numbered page of the previous issue is NOT known, therefore these pages cannot be numbered [156] and [157] (if they were known, then it is acceptable to infer the page number based on this knowledge as per Item 5.3.1) since these numbers may conflict with a previous issue, and it would not be appropriate to number them [4] and [5] because this groups them logically with the administrative content pages rather than the article pages.

If the source page number CANNOT be inferred AND there is no last known valid source page number, then key as follows:

nila where:

nil:

is an explicit prefix a:

is the sequential number for each “nil” page within the issue

(Exhibit 15)

As the source page number is used to identify each page image to be scanned, there is a requirement for the source page number to be unique.

The uniqueness of a source page number must be maintained, regardless of its appearance in the source material. Therefore, if duplicate source page numbers are present, the duplicate source page number is keyed with a sequential numeric -suffix appended.

EXAMPLE: There are two pages numbered in the source as “45”. The second occurrence of the page number is given a numeric suffix and keyed as “45-1”.

EXAMPLE There are a total of three page numbered in the source as “27”. Each occurrence after the first page “27”, is keyed with a numeric suffix. Therefore the second page “27” is keyed as “27-1” and the third keyed as “27-2”.

NOTE: If duplicate source page numbers are present, then an analysis of the page type is required. Often, differing page types may contain duplicate page numbers. Pages with article text should retain the original source page numbers. Pages identified as non-article should be appended with the sequential numeric suffix as described above.

OCR Specifications

OCR Engine

ASCII text files will be created for each article-level item using a high-quality OCR software package.

Editing

No editing will be performed on the OCR text as the good condition of the source material will generate sufficient character accuracy.

XML Specifications

Document Type Definition (DTD) and Schematron Article XML files will be created in accordance with the NISO JATS Journal Archiving and Interchange v1.0 DTD and must be valid according to it. As described in Section 7.0 , only a subset of elements and attributes from the DTD will be used, but some of the tagging details are more specific than can be tested with DTD. To this end, the NLM will provide a Schematron file which will be used to ensure the created XML conforms to the conventions described in Section 7.0 .

Font Changes and Super/Subscript

Font changes and super/subscript characters are captured using the emphasis elements described in Item 7.17. Font changes that are applied stylistically to an entire block of text (e.g. a title completely in bold) are NOT captured. Only font changes for strings within a block of text (e.g. a single italicized word in non-italicized title) are captured using the appropriate emphasis element.

Extended Characters

Extended characters (that is, characters not present on a US keyboard and that fall outside of the 7-bit printable ASCII character range) are replaced by their Unicode values expressed in numeric entity form (i.e., &#xHHH; where HHH is the hexadecimal Unicode value). (Exhibit 17) Unknown Characters

Special non-keyboard characters that do not have a valid extended character value are replaced with the “[unk]” placeholder string. (Exhibit 18) Handwritten Text

Handwritten text is ignored in the XML metadata; no special treatment is needed.

Illegible Text

Illegible text appearing in the XML metadata is replaced with the “[ill]” placeholder string. (Exhibit 19) Complex Data Structures

There are no provisions in the project DTD to capture tables, lists, complex equations and images in the abstract text. Therefore, these data structures appearing in the XML metadata are replaced with a placeholder string.

The placeholder string is formatted as follows:

[Object: see text] where Object is one of the following values:

Table Formula List Image

Filename Convention

Standard Conventions

Journal Code

The journal identifier is code assigned by the customer to each journal title. This code is used in the final deliverable directory structure, filename convention and article metadata.

Issue Identifier

The issue identifier is a 5-digit number for each issue. This number is unique only within a journal title. Issue identifiers are NOT assigned sequentially.

Sequential Page Number

The sequential page number is a 4-digit number assigned to each page image within an issue. The sequence includes all physical pages in the issue, including covers. The sequence does NOT include blank pages as such pages are excluded from conversion. The sequence within each issue will start at “0001”.

Page images are always assigned sequential page numbers according to the source sequence. However, in some cases, the page images will be re-sequenced in the PDF file as per specifications (see Item 6.11.3)

Sequential Illustration Identifier

The sequential illustration identifier is an alphabetic character assigned to each illustration appearing on the page. Identifiers are assigned to each illustration by strict sorting of the coordinates of the illustrations with respect to the full page image.

Article Identifier

The article identifier is a 4-digit sequential identifier assigned to each article within the issue. This number is unique only within a specific issue.

Article Sequence Identifier

The article sequence identifier is an optional alphabetic character assigned to an article if the article shares the same first page as another article. Identifiers are assigned to the illustration in “top-to-bottom, left-to-right” order.

This identifier is assigned to pages in the following cases:

· Page contains more than one article (see Item 5.2.1)

· Page contains an article and table of contents (see Item 5.2.2) or administrative material (see Item 5.2.3).

Full Page Images

The naming convention for full-page TIFF images (see Item 1.3.1) is as follows:

jidiiiii-pppp.tif where:

jid:

Journal code (see Item 5.6.1.1) iiiii:

Issue identifier (see Item 5.6.1.2) pppp:

Sequential page number (see Item 5.6.1.3) .tif:

explicit file extension

NOTE: For front covers (see Item 5.2.5), a suffix identifier, “-cov” is always appended. This is used to distinguish the cover page image from the table of contents page image.

Illustration Images

The naming convention for illustration images (see Item 1.3.2) is as follows:

jidiiiii-pppp-a.tif where:

jid:

Journal code (see Item 5.6.1.1) iiiii:

Issue identifier (see Item 5.6.1.2) pppp:

Sequential page number (see Item 5.6.1.3) a:

Sequential illustration identifier (see Item 5.6.1.4) .tif:

explicit file extension

PDF Files

The naming convention for PDF files (see Items 1.3.3 and 1.3.4) is as follows:

jidiiiii-ppppn.pdf where:

jid:

Journal code (see Item 5.6.1.1) iiiii:

Issue identifier (see Item 5.6.1.2) pppp:

Sequential page number of first article page (see Item 5.6.1.3) n:

Optional article sequence identifier (see Item 5.6.1.6) .pdf:

explicit file extension

XML Files

The naming convention for XML and XML “place-holder” files (see Items 1.3.6 and 1.3.7) is as follows:

jidiiiii-ppppn.xml where:

jid:

Journal code (see Item 5.6.1.1) iiiii:

Issue identifier (see Item 5.6.1.2) pppp:

Sequential page number of first article page (see Item 5.6.1.3) n:

Optional article sequence identifier (see Item 5.6.1.6) .xml:

explicit file extension

OCR Files

The naming convention for article-level text files (see Item 1.3.5) is as follows:

jidiiiii-ppppn.txt where:

jid:

Journal code (see Item 5.6.1.1) iiiii:

Issue identifier (see Item 5.6.1.2) pppp:

Sequential page number of first article page (see Item 5.6.1.3) n:

Optional article sequence identifier (see Item 5.6.1.6) .txt:

explicit file extension

Directory Structure

File/Directory Description jid\ Journal-level directory jid\iiiii Issue level directory jid\iiiii\adv\ Directory of advertisement files jid\iiiii\adm\ Directory of administrative content files jid\iiiii\art\ Directory of article files jid\iiiii\cov\ Directory of cover files jid\iiiii\toc\ Directory of TOC files

Articles Directory

This directory contains all files related to an individual article. The files for each article are contained within a subdirectory named as the corresponding article identifier.

Each article subdirectory will contain the following:

· Full-page images (see Item 5.6.2)

· Illustration images (see Item 5.6.3)

· Composite PDF file (see Item 5.6.4)

· XML file (see Item 5.6.5)

· OCR files (see Item 5.6.6)

Table of Contents Directory

This directory contains the tables of content-related files:

· Full-page images (see Item 5.6.2)

· Grayscale PDF file (see Item 5.6.4)

· OCR files (see Item 5.6.6) Administrative Content Directory

This directory contains the administrative content:

· Full-page images (see Item 5.6.2)

· Grayscale PDF file (see Item 5.6.4)

· OCR files (see Item 5.6.6) Advertisement Directory

This directory contains advertising-related content:

· Full-page images (see Item 5.6.2)

· Grayscale PDF file (see Item 5.6.4) Issue Covers Directory

This directory contains issue covers

· Full-page images (see Item 5.6.2)

· Grayscale or Full-Color PDF file (see Item 5.6.4)

· OCR files (see Item 5.6.6) Auxiliary Files

The following files will be contained within the journal directory:

· Batch-Level Checksum File…