Evaluation of Reference Management Software on Windows 3.1 (comparing Papyrus with ProCite, Reference Manager, Idealist, Endnote, GetARef, Citation 7)


Maggie Shapland, University of Bristol, 6 March 1998

Modified May 1999 concerning Reference manager network licencing.

Based on Windows 3.1 software.

  • Evaluation for Windows NT software (July 1999).
    compares Papyrus with Endnote, Pro-cite, Reference Manager, GetARef, Citation 7, Bibliographic, Biblioscape, Library Master, Scribe, Refs.
  • Research Toolbox evaluation (March 2000)

    Introduction
    Reference Management Software
    Systems considered
    Hardware requirements and interface
    Facilities checklist
     Database structure
    Limitations and Indexing
    Retrieval of documents and document sets
    Addition, deletion and modification of documents
    Formatting and output of retrieved documents
    Documentation and help facilities
    Summary Advantages and Disadvantages of Each System
    Conclusions
    Appendix
     Glossary of database-related terms

    Introduction

        The University of Bristol has had a site license for Papyrus for several years and offers courses and extensive support. There is a chest deal and most of the other Universities use it. However, it is felt that a new evaluation should be completed to satisfy those users who want a more windows-based package, provided they were prepared to pay for it themselves. This document aims to show the differences between Papyrus and the other systems. Papyrus would still be the principle bibliographic manager that we would recommend on the basis of its price and the fact that there will be a proper windows version sometime. The advantage of being able to recomend one other package would enable a deal with the suppliers and aid support problems

    Reference Management software

        Reference/bibliographic management systems are characterised by the ability to handle references of books, articles, journals etc (reference types) in the same database without having many fields defined that are only applicable to one type of reference, for example journal details are not relevent to books. They typically employ variable-length record structure, divided into a number of loosely organised components or fields. Fields have repeating values, eg multiple authors. This is true of text retrieval systems too. The structure of the database is often predefined unlike more general text retrieval packages and database packages. This should make the package easy to use since it has been tailored specially for a particular task but obviously loses the flexibility of a more general package.

        References are often batch loaded from external sources so a library of import formats needs to be available plus the ability to specify ones own, and modify existing ones since they have a tendency to change.

        References need to displayed in different ways depending on the publisher so again a library of export formats needs to be available plus the ability to specify ones own. One should be able to specify the fonts to be used, and support a wide range of printers. In addition one should be able to output to a word processor. The real power is when one can have a word processor document open at the same time as the reference management system, embed a pointer to a reference directly and produce a bibliographical list at the end of the document (ie citing).

        Comprehensive indexing mechanisms (typically at the word level) are used so that large texts can be rapidly searched for particular strings. Word punctuation is ignored. Typical search criteria may include:

        • proximity matching (search terms must be located within a certain distance of each other or within the same sentence or paragraph)
        • fuzzy or partial matching (when only a part of search term specified)
        • searching via an on-line thesaurus
        • boolean queries
        • being able to refine search criteria on the current query.

        The distinction between bibliographic managers and more general text retrieval systems such as Idealist is that bibliographic managers are solely used to handle lists of references and thus need to handle the many different import and export formats as specified by different publishers, including such author layout, and to provide its own character fonts without needing a word processor whereas a text retrieval system provides better search facilities.

        Relational database systems such as ORACLE, and Microsoft Access can be used for simple text retrievals. They do not generally support free text searching of the type considered here very well, author lists, journal lists, nor do they offer different character fonts and punctuation. Access and ORACLE memo fields can not be indexed for fast retrieval. Only ORACLE is available for both mainframe and micros. Access is easier to use and can handle large text fields. Both offer string functions. They have to be used in conjunction with a word processor and tagged import files are a problem.

        A brief glossary of database-related terms is given in the Appendix.

    Systems considered

    I used a Windows 3.1 system to evaluate the packages (but did try them later on Windows95).

    A data file consisting of 28 journal articles (size 95429 byte) in BIDS ISI format (see appendix) was used for testing. Even if the package could read tagged input, extra field information sometimes gets loaded in the wrong fields so it was a good test. A BIOSIS tagged file was tried since it has inconsistent indentation. A delimitered file was tried since users often want to load references lists that they have created using Word or Microsoft Access.

    Hardware requirements and interface

    Papyrus Endnote ProCite Ref Mgr Idealist Cit7 GRef
    version7.0.132.33.47.0237.13.2
    indiv price (exc VAT) s lic £199£200 £200£74/112 £105 £149
    disk store (inc formats) 5M 6.5M10M 8M300K2.4M 4.4M
    test db227K 46K? 65K n/a--
    memory470K5.5M8M 8M 640K4Mb 2M
    DOS-based yes nono no no no no
    intuitivefairly fairly fairlyfairly no fairlyfairly
    security pwdyes nononoyes no no

    Most of the evaluated packages were very expensive and none offered a cheap site license. Departments must be prepared to pay a lot of money if they want to switch from Papyrus.

    Papyrus

      • Cost.£700/ year for site licence, free to users. It can run on low spec machines and does not require much disk space, but can take advantage of windows facilities when run from Windows.
      • ***stores on average, 1500 references/megabyte. It will use upto 1Mb extended memory.
      • does not use expanded memory. Can have problems getting enough conventional memory
      • Being a DOS based program it has limited mouse usage.

    Endnote

      • offers various deals depending on number of users and manuals across all platforms. eg
        5 user license £825 with manuals, £775 without.
        10 users £1550 with manuals, £1450 without. 20 user £2800 with manuals, £2600 without.
        If a site wishes to increase the number of copies they may do so in minimum quantities based on the price given in the additional order. eg if the site has 50-99 licenses they may buy a minimum of 10
      • will not run with Norton Anti-virus, (but will with Dr Solomon's)

    Idealist

      • offers various deals depending on number of users, access, 16 bit or 32 bit eg 5 user 16 bit read-only license £298 (£398 read/write) with manuals, 32 bit read-only £448 (£598 read/write)

    ProCite

      • Cost £200 + VAT, 2-4 15% discount, 5+ 20%. 5% discount for educational establishments (Bilaney prices), £175 (RIS).
      • Upgrade to version 8 costs $99.95 RIS (plus shipping), £85 Bilaney (including shipping)
      • 5 user netpack (either concurrent or systems) $977.17 (RIS) plus shipping, £716 (Bilaney) including shipping
      • version 4 requires 10Mb disc space and 8Mb minimum memory

    Reference Manager

      • Cost £200 + VAT, 2-4 15% discount, 5+ 20%. 5% discount for educational establishments (Bilaney prices) £175 (RIS).
      • Upgrade to version 4 costs $99.95 RIS (plus shipping), £85 Bilaney (including shipping)
      • 5 user netpack (either concurrent or systems) $977.17 (RIS) plus shipping £716 (Bilaney) including shipping. Note the network edition is for named end-user names only. Other people using the software will be using it illegally and will get emailed about single-user licence contravention.
      • version 8 requires 10Mb disc space and 8Mb minimum memory
      • It does not support long filenames

    Citation 7

      • Cost £71 if ordered from web with machine documentation. 10 user license £320. 10% student discount

    GetARef

      • educational discount available, multi-packs and network pack

    Database structure

    The structure of the database is predefined This section shows the reference types supported, the number of fields per type and whether the user can define his own. For example, the Law Department may find that none of the reference types provided could be used to store case records

    Papyrus Endnote ProCite Ref Mgr Idealist Cit7 GRef
    no. ref types 8 16 28 35 2 38 3
    article 29 12 19 24 yes 12 8
    book 30 16 30 26 no 9 9
    chapter 32 18 21 28 yes 12 10
    conference =article 18 29 25 no 16 no
    map 20 12 25 23 no 11 no
    patent 28 15 29 26 no no no
    thesis 22 12 16 21 no 15 no
    quotation 16 no no no no no no
    personal =other(18) 9 15 18 no 13 no
    program =other 13 22 24 no no no
    report =other 12 29 24 no 13 no
    audiovisual =other 10 32 22 no no no
    artwork =other 11 22 20 no no no
    user-defined no yes no no yes no no
    user-definable fields 3 4 5 0 - 0 0
    major/minor keyword distinguished yes no no no no no no kw
    image datatype no no v4 link v8 link no no
    DDE no no no no yes no no

    Papyrus

      • provides less reference types, but the manual describes how to decide which reference type should be used since for example, conference proceedings are notoriously difficult to categorise.
      • distinguishes between major and minor keywords.

    Endnote

      • less fields in each reference type. Journal article omitted: Also print; affiliated address; series; supplement. All fields and reference types "soft-coded" and are fully editable

    Idealist

      • one had to define one's own reference types based on the format of the import record, which was a pain. Thus one ends up with reference types for Medline, BIOSIS, Embase etc which must make it difficult to distinguish between BIOSIS chapters and articles.

    ProCite

      • It provides 45 fields, and 28 reference types to rename and access some or all of these fields. Some of these fields can be renamed by the user but searches use the original names or column numbers, so user-defined field names are just boilerplate names.
      • less fields in each reference type. Journal article omitted: Also print; location, abbreviated journal, affiliated address; series; supplement, issue editor. Day and month included in date.
      • Extra fields were medium designator, connective phrase, translated title, language, availability and URL
      • direct interface to and from web enabled by URL field and express paste (vesion 3.4 has URL field but does not launch netscape unless database/open URL selected)

    Reference Manager

      • less fields in each reference type. 33 reference types. Journal article omitted: Also print; affiliated address; series; supplement.
      • enables direct interface to the web enabled by URL field (Version 8 only)

    Citation

      • less fields in reference type and not customisable. Journal article omitted: also print, accession number, location, address, field a etc, comments, issue title, editors. Had to use full type to get extra fields.

    GetARef

      • only had 3 reference types. Journal article omitted also print, accession number, location, address, comments, keywords, journal series, issue, supplement, day&month, issue title, issue editors

    Limitations and Indexing

    Papyrus Endnote ProCite Ref Mgr Idealist Cit7 GRef
    max Field size 8000 32000 32000 16000 64000 32000 255
    max ref size 16000 64000 32000*fld 16000 10000 32000 16383
    max field/ref 32 27 45* 33 128 25 10
    max ref/db 2 mill 32000 unlimited 65000 16300(16) 64000 32767
    max index length ? ? 32000 ? unlimited??
    max word/index term ? ? 255 255 66??
    max stop terms 0 0 600 0 unlimited 00
    max go terms n/a n/a 16000 0 n/an/an/a
    max sort fields 6 5 6 1 1 phys 2
    max authors/ref 100 100 ? 255 n/a 255char
    max keyword/ref 100 ? ? 255 n/a ? 0
    indexing redefinition
    thesaurus indep n/a no yes no no yes
    word punctuants no no no no yes no no
    synonyms no no no no yes no yes
    broad/narrow term no no no no broad no yes
    index list no no no no yes no-
    indexing by option author,year default auth,kw,title author,yr,kw,jrnlthesr?indx file

    Notes

        Word punctuants allow "don't" to be indexed as such, rather than "don" and "t"

        If the package has an independent thesaurus, it can be edited

    Citation7

      • each sort creates a new file, so the user must open the sorted file to get the display in the appropriate order

    GetARef

      • thesaurus only applies to indexed searches. Invoked by !word
      • index not updated automatically, and needs to be created by user

    Retrieval of documents and document sets

      ?
      PapyrusEndnoteProCiteRef MgrIdealistCit7GRef
      comparitive(<,>)yesyesyesnoyesnono
      boolean queryyes3 condyes****yesyes2 condyes****
      all field searchyesyesyesnoyesyesyes
      field searchyesyesyes*yesyesyesyes
      group searchyesyesyes**noyesno
      proximity searchnonononoyesnoyes
      left truncationnoyesyesnoyesyesyes
      right truncationyesyesyesnoyesyesyes
      wild cardyesnonoyesyesnono
      range searchyesyesyesyr,idyesnoyr
      empty fieldyesyesyesnononono
      SOUNDEXnonononoyesnono
      case distinguishnooptionoptionnooptionoptionno
      builtin thesaurusnonoyesnononono
      browse by arrow keyyesyesyesyesyesnoyes
      browse by scrollingnoyesyesyesyesyesno
      string functions0000000
      date comparisonsyesyes****yes***yr onlyyesyr onlyyr only
      search refinegroupsyesyesnoyesnoyes
      query hitnoeach searchyesyeseach paramyesnoyes
      save searchyesnoyesyesnonono
      word frequency listnonononononono
      jrnl,auth,kw freq listyesnoyesnononono
      save doc setsyesnoyestransfernoyesno
      list set defyesnononon/anono
      (),and,or,not setyesyesnonon/anono
      diff databasesyesyesyesyesyesyesyes
      multi-file searchnononoyesnoyes
      max search term??255?66?255

    Papyrus

      • empty field found by expression `not field=*'

    Endnote

      • **** Unless the year index is switched on, a search for 97 will find references from the 1970s and 1997
      • author search works differently depending whether the index is switched on or not. If switched on, must not include forenames or initials else will not find author, so using term list one must remember to remove the initials to do the search! unless index switched off

    ProCite

      • sort keys predetermined on author, title and date for a quick search.
      • * Field comparison rather odd. Each of the 45 fields has a field number which is prefixed by #, so to to check for text in a particular field, you do not use the field name but the field number. This is due to the fact that, depending which screen or workform you are using, the 45 fields often have different names.
      • ****Expanded Retrieval Operators (Less Than, Empty, Contains, etc.)
      • ** Searches on title, author and date automatically perform a group search on several fields.
      • *** since date fields are not validated and include the day and month, 26 jan 1999 will find a reference (date 1/26/99 and 26/1/99) as will 1999 but jan 1999 will not. Not possible to do >Jan 1999. Date sorts seem to work but dates are displayed as entered rather than in a consistent format

    Reference Manager

      • The first and secondary authors were distinguished so one had to search author1, author2 and author3
      • Reference list window shown as a split screen, with one part showing the full text of the highlighted list. References marked to cite or print.

    Citation

      • have to start search from top each time, as only searches down from current record

    GetARef

      • boolean operators must be in upper case
      • ****extra boolean operators XOR, ANY, ALL, NEAR, NONE
      • refining done by using LOCATED function eg LOCATED and dog
      • since there was nowhere to store month, could only look for year
      • no lookup lists of authors, journals to aid searching
      • confusing to do SO(Brain) rather than SO=Brain. The field name had to be in capitals else the search did not look in the field

    Addition, deletion and modification of documents

        As many validation procedures as possible should be performed at the time update is requested (before indexes are rebuilt). Index rebuilding should be unnecessary during normal updates. There should however be utilities to rebalance or repair corrupt or unbalanced indexes. Data should never need to be reloaded purely for tuning purposes.

    Papyrus Endnote ProCite Ref Mgr Idealist Cit7 GRef
    validity much no *no no no no
    range check no no no no no no
    thesaurus input no no yes no yes no no
    spellcheck no no no no yes yes no
    dup ref check yes *yes yes yes yes yes yes
    term lists 4 32 3 + create 9 6 0
    journal yes yes no abbrev yes n/a yes no
    keyword yes yes yes yes n/a yes no
    author yes yes yes **yes n/a yes no
    glossary yes can create can create no yes no no
    count yes no yes no no no no
    auto update yes no yes yes n/a no no
    auto keyword scan no no no yes n/a no no
    synonym editor jrnl jrnl jrnl auth,kw,jrnl jrnl jrnl
    repair indexes yes souvenir yes yes yes no yes
    global editorkw,jrnl,authany fldlimitedkw,jrnl,auth yes no no
    batch loading formats 190 27 360 230 16 100
    free update yes yes no no ? yes yes
    mailmerge yes define yes no yes no no
    labelled(tag) yes yes extra yes yes yes yes
    usr-delimit yes define yes no no no no
    fixed format no no no no no no no
    MARC no yesno no no no no
    userdefin yes yes Bibliolink yes yes hard no
    Papyrus yes .flb delim .flb yes no .flb
    BIDS formats 18 1 0 3 Leeds 1 1
    debug IP format some no no no no no
    batch reject file yes no no no no yes
    2 digit year entry yyyy yy *yy yyyy yy yy checks
    extended ascii yes yes yes *yes+greek no yes no
    OLE link no no v4 only v8 only no no

    Notes

        Most external references come as tagged input in various styles hence the need to be able to define an appropriate format. Surprisingly, BIDS formats varied as to which fields to ignore, and which field information was stored in which field! Surely this should be consistent. It is also surprising on the variance of how many BIDS formats should be needed

        All hard to debug import formats. Papyrus just apologises but at least gives helpful logs and parsing display. Reference Manager gives a misleading message. All except Papyrus tell you what has been imported, but not what it has not. It would be really nice if the packages gave a clue as to which line they did not like

        To import mailmerge(word documents) or delimitered files (one record per line)- one would have to try to import the file into Microsoft Access and then write a report to output it in tagged form which is very cumbersome if a delimitered format is not available

    Papyrus

      • Manchester have provided 18 BIDS formats for Compendex, Embase, IBSS, Inside Information, ISI, ISTP
      • required all fields to be surrounded by quotes for import if mailmerge format was used
      • gave the most useful information when importing files ie new journals, keywords, authors, capitalisation information, and one could at least watch it scan the text. It also held duplicates and rejects in other files for checking purposes
      • checks authors names etc and if typed in in lower case, whether this was correct
      • the most flexible package since one could specify the layout of the tags. There was no control over indexing. Separate files were created to hold rejects and duplicates but the only assistance to check why import failed was to watch parsing with Caps lock on
      • will enter a date as yyyy if only the last 2 digits are supplied,

    Endnote

      • To import delimitered files, Endnote provided tab delimitered format which also needed authors separated by ; before import. It was easy to provide a user-defined delimitered format
      • Import options were available for Refer, BibIX, ProCite, Reference Manager.
      • Only one BIDS format. Imported Day and month into issue, RF into keyword (ignored in Papyrus), CR (cited ref into Papyrus comment), NA (address) were all ignored
      • Term lists not updated automatically which was confusing
      • Specifying import formats very straightforward. I created one for BIOSIS with no problem. The biggest problem was that no log was kept of new Journals, keywords, and any omitted data so although data was imported it was sometimes incomplete. It was also difficult to decide where to put the extra information such as address, and minor keywords, so the Notes field got filled up with different kinds of information with no way of separating it. The BIOSIS file I tested out (which inconsistently does not have indented abstract information even though all other continuation lines are indented) only entered the first line of the abstract starting in column 10 and ignored the rest.
      • very good at interpreting authors names
      • makes no attempt to add a century indicator, nor is it possible to add a century indicator later except manually since a text field
      • can create a term list for each field
      • no good at detecting duplicate references once they were stored
      • built-in import options for Refer, BibIx, ProCite, Reference Manager. conversion utilities for Papyrus, Notebook, Notebuilder, REF-11, Publish, SRS, REFLIST, Bookends, Citation
      • EndNote 3.0 beta will search MEDLINE over the Internet, store the references in a searchable database, and automate the process of citation.
      • A Papyrus .flb file is obtainable from ftp.niles.com to export from Papyrus. This file is then imported as Refer/BIBIX format
      • formats obtainable free from ftp server

    ProCite

      • ProCite delimitered importing was difficult. Required 45 fields per reference, all fields to be surrounded by quotes for import if mailmerge format was used
      • Trying to define a custom delimitered file format to reduce the number of fields did not allow one to specify the order of the fields making it totally impossible to import files originally created using Word
      • Tagged formats required Biblio-Link to be purchased which do not allow the user to modify them. Version 4 was able to import tagged files without BiblioLink and now includes Biblio-Link II free which now allows format modification
      • Technical support is free- they can advise on how to edit your own format. One may use the database conversion service provided by the company programming staff at $250 per conversion and $.05 per reference- more if one is in a rush (normally 4-8 weeks)
      • If year is not included, it is not added automatically
      • *Date fields do not have to be entered consistently, and include the day and month. The only problem is that the international settings are ignored so 26/1/99 and 1/26/99 are both accepted (1/9/99 is 9 Jan) as is April 99 etc
      • multiple authors entry was messy since one had to separate each by 4 slashes. Book author and journal author were distinguished
      • global edit was very limited in that it could not if the word had a different font. It could only add prefixes and suffixes, not replace words or authors
      • built-in import options for Reference Manager
      • web form in version 4 to enter data.
      • titles and journals distinguished in version 4
      • only full and abbreviated name information could be held about journals
      • To transfer from Papyrus, create a format to create a 43 field comma-delimitered file
      • input format editor BiblioLink needs to be purchased separately if using Version 3.4
      • formats obtainable from web free. One can pay $79.95 per annum maintenance fee to be automatically sent new import filters, program fixes an minor upgrades

    Reference Manager

      • conversion utility to import from Endnote, Papyrus, ProCite. Papyrus flb file must be provided and then the resulting file imported using the Papyrus format (not described in help).
      • updated formats from the web are available free Maintenance including new formats one must pay $79.95 per year
      • BIDS formats provided are Tagged, Embase, ISTP
      • cannot define or modify formats- unless order capture editor from suppliers
      • Technical support is free- they can advise on how to edit your own format. One may use the database conversion service provided by the company programming staff at $250 per conversion and $.05 per reference- more if one is in a rush (normally 4-8 weeks)
      • could scan titles and notes so that keywords already in the term list could be added as keywords to relevent references
      • it appeared to be quite easy to import a tagged field into the wrong Reference Manager field due to it being generic. In fact, the BIDS tagged format put address into the notes field, since there is no address in the journal reference type definition (could relabel user-defined1 as address but then one is out of sync with other users). The BIDS format also put abstracts into the Notes field!
      • Authors initials can import as part of the last name if delimiters not specified correctly
      • if one entered a new author name in lowercase, it stayed in lowercase with no comment, but if already known, it was converted to mixed case
      • get the standard message "reference too large to import" if format does not match with no more help
      • cannot import non-tagged files- one has to use the database conversion service provided by the company programming staff at $250 per conversion and $.05 per reference- more if one is in a rush (normally 4-8 weeks). Help is available free
      • XX was imported to a field if there were unexpected carriage returns in the file
      • *Extended characters, and Greek did not work very well- there were only 5 letters in the greek alphabet, and è੾¼ came over as èàc_¼ using the character map button

      • **entering a new author in lower case, did not cause automatic case conversion. Only entering a known author would case not be distinguished. It was also necessary to change an option to allow fullnames, despite this, Reference Manager still asked to initials to be separated by dots

      • version 8 included the ability to store the path to an OLE object. A click launches the relevent application to display the object
      • version 8 enabled web URL, email addresses an medium to be embedded to enable direct access to the web
      • Could not get hold of input format editor from suppliers
      • keyword synonym and author synonym editor allowed alternative names to aid searching

    Idealist

      • I made no attempt to import since it was too hard to set up.
      • Leeds have done alot of work with BIDS

    Citation

      • spell check even checked authors and initials!
      • term lists held in external files which are edited with Notepad
      • Format editor looked horrendous, very program-orienated
      • import files "converted" into temporary file to check wich is then added to the main datafile. Addfile gave a pagefault in kernel (Windows95)
      • import instructions for Endnote, Notebook, ProCite, Reference Manager, Papyrus. Have to get Reference Manager flb to get data from Papyrus in suitable format
      • bids format put first author and date as unique identifier (access key)
      • when new journals were input, one had to open the journal file and add new journals manually so that abbreviations could be added. There did not seem to be a tool to add journals from a datafile that were not in the datafile. There was also a publisher abbreviation file

    GetARef

      • auto-detected BIDS format!
      • Import facilties provided for EndNote, Reference Manager and Papyrus, ProCite, Bibliographer, RefII
      • To import from Papyrus have to export using PAPX format. Keywords are placed at the end of the abstract field
      • formats obtainable free from ftp server
      • comes with journal abbreviations for 17,000 scientific names
      • Extended characters obtainable from Windows Character Map

    Formatting and output of retrieved documents

    PapyrusEndnoteProCiteRef MgrIdealistCit7GRef
    format layout
    default 2 . 7 5
    predefined 126 300 30 192 . 100 9
    page headings yes ref hdg no no . yes no
    duplicate ctrl yes no yes yes . yes no
    sort any fld yes yes yr,auth,id . newfile yr,auth
    margins yes no yes no . no no
    indentation yes messy yes no control . yes yes
    page length yes no yes no control . no no
    page nos yes no yes no control . no no
    underlining yes yes yes yes . yes yes
    bold,italic yes yes yes yes . yes yes
    super/subscript yes yes yes yes . yes yes
    capitalisation yes no all caps all caps . yes yes
    change fonts no yes yes yes . no yes
    alter author format yes yes yes yes . yes yes
    et al yes yes yes yes . yes yes
    save format spec copy yes yes copy . yes yes
    printer op yes . yes yes . no yes
    rtf op no yes no no . no yes
    ASCII file op yes yes yes yes . no yes
    spec printer info yes . yes yes . no yes
    Word interface 1-7 1-8 1-8 1-6 no 6-8 6-8
    Word add-in no yes yes yes no yes yes
    cite directly into word yes yes yes yes no yes no
    web output userdef** yes NetCite v8 . yes userdef

    Notes

        If the fonts cannot be changed, then a word processor has to be used. The windows packages allowed one to highlight fields or text.

        Most packages could only cite with Word6 documents, requiring documents to be saved as Word 6 if using a later version so did not support later versions properly

        Word Add-in made citing easier since did search for reference and bibliographical generation from Word

    Papyrus

      • **possible to define an HTML format for a list but not in-text citation.
      • also supports Word Perfect, PC-write, Word-star, xywrite/Signature, Tex, latex, Chi writer, Ami Pro, ASCII, ASCII with line breaks
      • citation prefixes and suffixes can not be specified
      • Papyrus lists default to the Word default style setup
      • latest version supposed to support Word 7. If the text extract fails then the document must be saved as a Word 6 document

    Endnote

      • provided an HTML format for both a list and an in-text citation.
      • installs add-ins (16 bit for Word 6, 32 bit for later) into the tools menu in Word 6 onwards (and Word Perfect) to format and cite. (Endnote Paper commands not used with Word 8 unless one saves Word 8 documents as Word 6/7 or rich text format before scanning.) To edit a Word 8 document with any other version of Word and intend to reformat citations, unformat the citations before transferring the document.
      • also supports Word Perfect, RTF, Ami Pro, ASCII
      • allowed citing from different databases provided they were all open
      • reference and citation prefixes and suffixes can be specified

    ProCite

      • also supports WordPerfect
      • installs add-ins into the tools menu in Word 6, 7 and 8 (and Word Perfect) to format and cite
      • v3.4 supports Word 6, 7, 8. v4 supports Word 7 and 8
      • can only list 6 fields (recno, author, title, date, keyword) when scrolling

    Reference Manager

      • also supports WordPerfect, Word Pro, Ami Pro, Wordstar
      • option to create add-in in Word
      • Express Paste does not work with Word 7
      • could not switch off the pagenumbers from the bibliographic list or change page layout
      • v7 compatible with Word 6, 7 since it works with 16-bit word processors. v8 compatible with Word 7 and 8. Version 8 is written specifically for Windows 95 and NT so it is designed to work with Word 8 and utilises many NT features such as right mouse support. I you want to use Word 8, you must use Reference Manager version 8
      • If using v7 with Word 8, save the document as an RTF file, change word processor to ascii in Reference Manager, change the citation delimiters. If the journal format includes any special text style, the bibliography will have to be MANUALLY edited. The RTF format only retains the font styles of the original document, not the in-text citations or bibliography list. (version 8 works fine with Word 8)
      • Does not recognise a file that has been fast-saved as having been created by Word, so the allow fast saves must be disabled

    Citation

      • Also supports WordPerfect
      • citing needed the user to identify each reference by user-specified unique "access key" rather then automatically generated record number
      • citing (add-in in Word only switched back to Citation7), required the user to put {} round the pasted access key. Generate citations from doc then correctly added the cites in the appropriate style, but unless the reference list box was checked no list was generated. The list had a different font and size to the rest of the document. Generate Biblography confusingly created a list of all references
      • sorted output was placed in a new datafile, as was selected data, so a plethora of files was created (sorted and selected required 2 files).
      • the only way to print was to generate a bibliographic list where Word was already open so the output went directly into the current document. If one output to file, Word 6 could not open it since "file is not a Word Perfect document". The other option was to output to the clipboard

    GetARef

      • When citing, can only scan rtf, ansi, ascii, Word Perfect, WordStar, Ventura Publisher Sprint.
      • Add-in merely opens GetARef. To cite one must use GarDoc program

    Documentation and help facilities

      PapyrusEndnoteProCiteRef MgrIdealistCit7GRef
      Documentation
      User manual1111110
      Reference manual1no1noyesno
      Tutorialyesyesyesyesyesyesyes
      Readabilitygoodgoodgoodgoodaverageaverageaverage
      Indexgoodgood..poorpoorpoor
      Help
      keywordnoyesyesyesyeslimitedyes
      in contextsomenoyesyes.nono

    Summary of Advantages and Disadvantages of Each System

    Papyrus

        Advantages

          • only £700 for a site licence
          • has a versatile bulk-import and bibliographic creation facilities
          • good interface to word-processors
          • good search and sort capability
          • most "helpful" of all the packages for importing since logs kept
          • journal, author and keyword counts
          • good free support from vendor and other Universities
          • runs on Mac and Windows. Mac version of PAPYRUS is available for free downloading from Web site (late beta version now, limited/demo version later), and has many new and greatly-improved features compared to Version 7.0.
          • checks keywords, authors and journals on input against existing values
          • can import mailmerge file and define user-delimitered files
          • able to distinguish between major and minor keywords

        Disadvantages

          • needs full windows facilities, cutting and pasting limited, no scrolling
          • brief on-line help
          • handles eight reference types. It cannot easily store references to manuscript collections.
          • citation is a 2 step process requiring switching between applications
          • cannot define html citation style, only for bibliographic list html
          • can run on network provided all users read only

    Endnote

        Advantages

          • good on-line help and 15 pre-defined reference types and 300 pre-defined output reference styles,
          • input and output formats can easily be modified by the user.
          • Endnote less fussy than Papyrus when importing references especially with author names where smart parsing could be used
          • good support from vendor
          • works well with Word 8
          • HTML style to create HTML documents
          • runs on Mac and Windows with same features (unlike ProCite and Reference Manager)
          • can import mailmerge file and define user-delimitered files

        Disadvantages

          • author search confusing, and limited retrieval operators
          • term lists not updated automatically
          • no term list counts
          • were far less fields for each reference type than Papyrus
          • fell over if file handles exceeded necessitating reboot
          • Must deactivate Norton anti-virus before installation to avoid freezing
          • no link to OLE compliant files
          • can run on network provided all users read only (but no special license needed)
          • no direct interface to web

    Idealist

        Advantages

          • good search (the only one to have proximity searching) and cross-refence facilities

        Disadvantages

          • limited import and export capability
          • cannot generate bibliographies from citations in a text.
          • not intuitive and does not offer many bibliographic manager facilties
          • only about 2 predefined reference types

    Reference Manager

        Advantages

          • Available for both MAC and PC
          • large number of pre-defined reference types.
          • Five user definable fields available, and able to rename other field labels
          • wide range of citations from biomedical reference sources can be imported
          • direct interface to web (version 8 only)
          • Windows 95/NT version with 32-bit, property sheets, common dialogs, OLE, long file names, small icons, and right mouse support)
          • can run on network provided only 1 user writing, other users (max 4) read only (with special license)
          • Formats may be downloaded from the web

        Disadvantages

          • Must run version 7 if one has 16 bit system, version 8 with a 32 bit system which makes support difficult.
          • The journal reference type was very limited since it did not include journal series, supplement, day and month, issue title, editors of issue, address, location, or distinguish between major and minor keywords. In fact, importing BIDS data using the BIDS format put address and abstract into the notes field, and BIDS notes and day and month got discarded.
          • no user-definable fields for own use
          • BIOSIS import formats did not match the file that I had.
          • not able to try out the capture (import format) editor but it is free
          • The suppliers will create formats for a fee (but offers advice for free).
          • very confusing to distinguish between primary and secondary authors.
          • no term list counts
          • The extended character set did not work very well.
          • Mac version did not have the same features as the NT version
          • can not import mailmerge file (useful for users who created database with Word or Access)

    ProCite

        Advantages

          • Available for both MAC and PC
          • has an attractive interface with on-line help
          • many CD-ROM databases offer output in ProCite format.
          • Windows 95/NT version with 32-bit, property sheets, common dialogs, OLE, long file names, small icons, and right mouse support)
          • can import mailmerge file (useful for users who created database with Word or Access)
          • delimitered file import with 43 fields totally impractical
          • term list counts
          • direct interface to web
          • Formats may be downloaded from the web and then edited.
          • can run on network provided only 1 user writing, other users (max 4) read only (with special license)

          Disadvantages

          • expensive, especially when one has to pay extra for import formats, and one cannot specify ones own. The NT version rectifies this so hidden extras no longer a problem.
          • Mac version did not have the same features as the NT version. The latest version 4.03 has the same features in Mac and Windows versions.

    Citation 7

        Advantages

          • cheap

        Disadvantages

          • no Mac version
          • hard to create import formats, limited number available
          • had to scan rtf file to format citation document
          • no link to OLE compliant files

    GetARef

          Advantages

          • Intuitive import, telling user which format the file was in
          • cheap

          Disadvantages

          • no Mac version (but can work on Power Mac emulating windows)
          • limited number of reference types
          • reference types omitted keywords and comments fields amongst others with no spare fields to store extra information
          • confusing to run different programs for each function-GetARef to edit and retrieve, GarConv to import, GarComp to compare, GarList to create reference lists, GarForm to create formats, GarDoc to complete a text document and GarSort to sort a reference file
          • had to scan rtf file to format citation document
          • could not write own import formats
          • no link to OLE compliant files
          • not one-step citation

    Summary (* indicates important features)

    Import: Output:
    - Papyrus Endnote ProCite Ref Mgr Idealist Cit7 GRef
    DOS v7.0.13 yes v2.2 6.02 v3 no no
    *MACINTOSH v8(dev) v2.3 v2.1 v2.51 v3 no no
    Windows3.1 works v2.3,32bit v3.4 v7 v3 v7.1 3.2
    Windows95/NT works v2.3,32bit v4 v8 yes v7.1 3.2
    over network yesyesyesyesyes
    backup facility yes no no no
    Var fields yes yes yes no yes
    vocab list yes no yes yes yes
    Intuitive fairly fairly fairly fairly no fairly fairly
    Reference Types 8 16 26 33 2 38 3
    Reference fields(max) 32 27 45 33 128 25 10
    user-def no yes yes no yes
    ext char set yes yes yes limited yes
    Link to OLE Files no no yes yes yes no
    extra boolean operators no yes no no yes
    Multiple Database Ops no yes yes best no
    Editing
    validation aut,kw,jl no no no spell spell
    use term list yes yes yes yes no yes
    global some some limited some yes no
    tagged yes yes extra yes yes
    library yes yes extra yes no
    * user-def delimitered yes yes no no yes no no
    * style editor Q+A intuitive now free now free no hard not imp
    dup ref check yes poor yes tailor yes yes
    Keyword Scan of Text No no Yes no no no
    * style editor Q+A visual visual visual no hard visual
    * Word 6/7/8 save6 yes yes v7no,v8 yes no 6-8 6-8(rtf)
    * one-step cite no yes yes yes no no no
    HTML format can define yes yes yes no yes
    web link no not yet yes yes no no no
    Journal abbrev yes yes yes yes no yes yes
    year 2000 compliant yes no** mess yes no* no**

            *Year was a text field rather than a date field (sorted out in next version of Endlink)

    Conclusions

            It is very difficult to arrive at a conclusion since no one package was good at everything, and all are going to cost departments a great deal of money. If only Papyrus had a proper windows interface there would be no need to recomend another package since it is cheap and powerful. One also needs to standardise on which fields one imports BIDS information into since each supplier had his own ideas. One could export Papyrus databases into all the packages, but principally this is done by providing the correct output format for Papyrus. On the basis that most of the support that the users require is for importing external references, I consider that Endnote is the best package since it provides the most intuitive, flexible facilities. Endnote is in the process of bringing out a new version so hopefully will incorporate better retrieval facilities

          • Endnote was good at importing, but retrieval was confusing, term lists were not updated automatically and network facilities were limited
          • ProCite and Reference Manager offered interface to the web and OLE facilities but were bad at importing, and had to pay extra for formats every year. They have now redressed this problem so the hidden extras do not now hit you at every corner, but they still charge for creating formats.
          • Reference Manager had synonym editors for authors and keywords to improve searches version 7 did not work well with Word 8
          • ProCite had good boolean operators but date handling was totally confusing and unacceptable. Did not provide a format for BIDS
          • Idealist did not offer reference management facilities so can not be considered
          • Citation7 was cheap but did not work on a Mac and formats hard to define. Had many reference types but no user-definable fields
          • GetARef was cheap and good searching but one could not define ones own import definitions. Had very limited reference types with no keywords. Had to generate cited documents with rtf format

    Appendix Glossary of database-related terms

            Authority list
            A list of phrases or words that may be used as a look-up table for indexing, control of document vocabulary, search list. Lacks thesaurus facilities of being able to expand search to broader/narrow subjects.

            BIDS format
            TI: title of reference AU: surname_initials, surname_initials,... NA: address (each line terminated by /) address (each line terminated by /) JN: Journal date as year vol.volume NO.no PP.page-page DT: document type PA: Cited patents CR: Cited references RF: Research fronts

            Compact Cambridge format

               UI: UNIQUE IDENTIFIER
	reference identifier
	AU: AUTHOR
	surname initials; surname initials....
	TI: TITLE
	title of reference terminated by '.'
	SO: SOURCE
	Journal; date as year month; vol(issue); P pages
	AB: ABSTRACT
	abstract. ABSTRACT.

            Database
            A collection of interrelated data typically held in a series of relations which are manipulated by SQL.

            Data dictionary
            Data held about the data structure itself.

            DDE
            Dynamic Data exchange- enables data to be swapped with other applications

            Document
            An item of information in full-text form.

            flb
            The file type of a Papyrus file containing input and output formats

            Go word list
            A word list used to find documents and validate new words (an index)

            MARC
            MAchine-Readable Catalogue, a standard format for interchange of data between libraries. Each tagged line may contain any number of subfields. Reliance on numeric tags

            Punctuation file
            Defines bibliographic style sheets to format a bibliographic record.

            SDI capability
            Single Document Interface. This would enable users to search a database using the same criteria at regular intervals, retrieving only documents that have been added to the database since the last search was made.

            SGML
            Standard Generalized Markup Language.

            SOUNDEX
            A phonetic matching scheme.

            SQL
            Structured Query Language, the standard data manipulation language for relational databases.

            Stop word list
            A common word list eg 'and', 'or', 'the' which are not used to find documents. Used to save space by not being added to index

            Thesaurus
            A structured dictionary controlling the use of vocabulary. An integrated thesaurus is felt to be an essential component of the system. It should be possible selectively to control terms entering the database and to maintain authority files. Index terms should be retrievable and displayable in the same way as documents, preferably using the same commands and utilities.

            Truncation
            Example: If search for 'car' and retrieve 'cartoon'

    General list of Requirements

    • good sorting and retrieval capabilities
    • ability to handle sizable datafiles
    • friendly, consistent user interface
    • Windows interface
    • Ability to handle large text fields
    • Ability to import from a variety of sources
    • Ability to define import format (tags)
    • Ability to output to a variety of sources
    • Ability to specify italics, bold, underline, fonts and capitals
    • Ability to export to a word processing package
    • Versions of Word supported
    • Ability to change format of database
    • Ability to handle articles, books, etc in same database
    • simple and readable user manuals
    • training aids
    • context sensitive help
    • switchable case sensitivity
    • software tools to check for consistencies
    • keyword and thesaurus tools
    • cost and value
    • licencing agreements
    • networking facilities
    • machine range
    • web interface