Introduction

It is known that routine updating with new documents is a labour-consuming work in maintaining of bibliographic and other information retrieval systems (IRS). The Thermophysical center in the Institute for High Temperatures of Russian Academy of Science developed a technique for efficient organization of the input information to the IRS system CDS/ISIS for Windows (ISIS). The Thermophysical center for decades have used ISIS of different versins, developed by UNESCO, for its main database THERMAL on properties of substances.

ISIS has two means to enter new information:

• with system interface, consistently filling fields of the enter form by text from a document;
• to import a file with group of documents in format of ISO 2709 standard for data exchange.

The second way is much more effective, but it is necessary that document file would be prepared by some way, for example, it can be exported in ISO format from other IRS that has function to do such conversion.

It is necessary to consider one more important circumstance that appeared nowadays and connected with global Internet. Within it many documents (articles, books, materials from periodicals, etc.), that should be included in databases, are became accessible in an electronic format. It allows to reduce manual typing of texts in fields of the system interface and considerably to diminish expenses on data input.

The ISIS system interface can be used in this case too by copying line to line from the electronic document via clipboard and then input the ready document in the usual way like hand typed.

But it is possible also to mark in a special way the text of electronic document and convert this information to data exchange format ISO 2709. The records prepared thus from different sessions and/or from various staff operators can be grouped in a file and added to a DB during a single session.

Also it is not a problem now to use the same technique when a relevant document is available as a hard paper. With a scanner and recognition text program it can be easily transformed into electronic form. In some cases only necessary part of the document (authors, title, abstract, and etc.) can be scanned for input to a database. After that transformation the procedure mentioned above can be applied to get records in ISO 2709 format.

As an example, I show now some details of the approach that was applied in Thermophysical Center (TPC) of the Institute for High Temperatures in Moscow, Russia.

It is easily understood that a text marking scheme must correspond to DB structure in ISIS. So, having aim to input into a DB, one must use the same tags in marking procedure, that may be different in every ISIS data base. Such approach has been developed, and program IsoWin (Beta-version) making transformation of marked record in exchange format ISO 2709 was created. The procedure is illustrated by examples of a real DB on thermophysical properties of pure substances TERMAL.

The scheme of text marking

Letters on the right row (indexes of subfields) show marks of repeating records for those fields.

As one can see the rules of text marking for ISIS databases are simple:

• The text which have to be converted to ISO 2709 format must begin with the symbol @ and finish with line of symbols $$$. Everything, that is outside of these symbols, is excluded from transformation and can be used for notes and instructions for work both in the beginning of the text and between documents.
• The information, which have to be entered into a field of DB, begins with tag of this field (three figures in ISIS) and ends with a symbol tilde ~.
• In case of several repeatable subfields they are marked by a symbol ^ with the subsequent letter (the subfield index FDT).
• If a field contains only information for one subfield, its index may be omitted.
• It is not necessary to start new field on a new line. Some fields can be put in the same line of the text.
• Sequence of fields in a record may be arbitrary.

Apparently, these rules are obvious. In an example on the Fig.1 the basic part of record in English was taken from an external electronic source (Internet, fields 001-006). The other part (search terms, names of substances and properties in accordance with thesaurus of DB TERMAL in Russian, and others) have been added by expert-reviewer on analysis of the publication content in whole.

Program IsoWin

This program fulfil transformation of the marked text to exchange format ISO 2709. IsoWin runs in Microsoft Windows95 and above. The program consists of two basic functional parts ¾ own text editor for text marking and the converter of the marked text. Generally speaking, one may use any other editor which keeps the text in a file precisely the same as it is visible in a window of this editor. An example of suitable editor ¾ Microsoft Notebook. There should be no hidden symbols in the marked file that operate kind and size of fonts, some tabulation, etc., except for commands "go to a new line " and "move the carriage ".

In this Beta-version editor IsoWin fulfils all operations typical for text editors, but for simplification of work with external sources it has following additional features:

• The tabulation codes in documents are ignored in final processing.
• Hyphens thad divide words on two lines are eliminated at converting, i.e. they do not need to be edited manually. The line of blanks is reduced to one.

• The length of a line in text should not exceed 80 symbols.
• The length of one field in the document should not exceed 30 Kb.
• The length of the file entered into a window of the editor, also should not exceed about 30 Kb. At converting to exchange format file become larger and can exceed admissible limits. The enlargement is more significant for a file with many short documents. In case of memory shortage the file should be segmented and then collected after results of converting are received.

After the marked file is prepared in the editor window or open ready from an other source, its converting to exchange format is run by command GetISO (the button on the panel or the command in menu Make). The result is visible in the same editor window. The structure of exchange format is not discussed here. I’ll note only, that record of each document begins with long sequence of the figures that codes its structure. So, it is possible to allocate visually separate records and to receive some understanding on of the file content. Generally speaking, minimal editing of result is possible in mode "replace" of the keyboard without change of field length.

As it was said above, texts in internal records of databases in ISIS have DOS coding (OEM), but nowadays they are prepared or downloaded practically always in Windows (code ANSI). Therefore, converting in an exchange format can be carried out by the program in two modes ¾ with transformation of the coding from Windows into DOS (ANSI to OEM) and without such transformation. The second mode is used in the case when the initial file is prepared in OEM coding. It is necessary to note, that though transformation ANSI to OEM is indicated in ISIS, in case of Russian its results are inappropriate. Data import in ISIS needs file in OEM coding. Probably, the same occurs with other languages also. In this version transformation ANSI to OEM fulfils only for Russian with own accordant table. Russian letters in an input file, that was coded in OEM by any other editor, in IsoWin editor look unreadable and are not fit for editing. The usual operating mode is typing and text editing in standard Windows mode, and then converting of result into OEM.

• collect SWin string by typing all letters that differ from usual English ones;
• transform it to DOS (OEM) with appropriate local editor;
• get Sdos string by opening the transformation in any local Windows editor.

Such algorithm allows to make localization of program IsoWin simply enough. So, such localization does not change its English interface, but it is not an inconvenience because it is common in computer world.

Certainly, program IsoWin can be used also for other databases that accept information in exchange format ISO 2709 in different coding.

For many information services IsoWin may practically become the basic tool of workplace for an operator, working on data preparation to input in a DB.