I realise that in criticising the recent IFLA Draft Guidelines to OPAC Display produced by Martha Yee I am criticising a document that is thorough, competently produced and which pinpoints a number of problems in the design of most OPACs in use today. It has also been enthusiastically received by a number of commentators. However, I am doing so because as it is based on OPAC design as used in most vendor systems operating today it ignores the potential for change brought about by a number of technical developments. These will affect the way in which OPACs are designed and even more important the way in which they are used. As these developments are likely to escalate in the near future the Guidelines are likely to become out of date very quickly. The factors that are likely to bring this about are:-
The Internet is the most important of these developments and new innovations and improvements are being introduced so fast that it is often almost impossible to keep up with them. What is certain is that these developments will profoundly affect library catalogues. It is now possible to access probably hundreds and possibly thousands of library catalogues through the Internet. It is also possible to search groups of these catalogues using the Z39.50 protocol so that a number can be searched at the same time. Although Z39.50 is not perfect it is developing and improved versions are appearing at regular intervals.
The growth in the use of the Internet has meant that many people now have the basic skills to develop simple Web pages using HTML. However, although these pages may be both attractive and functional, authors would like to make them more dynamic and their content more interactive. For example they would like to make the user mouse and keyboard actions more reactive, and provide configurable pages and the local processing of data. Most of these are beyond the capabilities of HTML. The development of JavaScript provides a simple extension to HTML which is easy to learn and can quickly produce sophisticated and effective results. Over the last two years the Java programming language has come into widespread use on the Internet and many sites have Java applets. which enable visitors to move around and exploit these sites quicker and more effectively. BOPAC2 uses these facilities to open up new possibilities of OPAC display. The essential difference between BOPAC2 and the sort of OPACs that the Guidelines are designed to cater for is that once a retrieval has been obtained and the Java applet has been loaded the user is able to move very quickly between a large number of different displays. Each display can be scrolled through quickly and it can be broken down in a number of different ways:- initially by author or title and then by other options like publisher and format. These can then be examined in detail. This is impossible in the traditional OPAC. Even though BOPAC2 is a product of a research project it remains on the Internet as a working tool and is still being used all over the world.
One of the routine things that BOPAC2 does is to arrange the initial retrieval in some useful breakdown. To see this for any retrieval press the select all button and then the look at these button. We believe that it is better than any of the examples given in the Guidelines. Although BOPAC2 concentrated on known item author title searches we believe that with further funding for research and development this could be improved dramatically. In some cases a subject approach is quite impressive. At this stage we have only thought about it and tried some simulations. For example sort the retrieval by the first 650a field. If members would like to test this approach I suggest they look at the thirty retrievals for 'angels flight' from MELVYL using 'title contains' as the search term and then press F2 to get the MARC format display. Scrolling through this format looking at the 650a fields provides a good simulation of how this display would be grouped.
I think that the points that I have made are important and demonstrate that the structure of OPACs can change and that it is dangerous to base any Guidelines on a technology which is likely to be superseded. The evidence seems to point to much larger retrievals based on broader search criteria. These retrievals will be reorganised and searched very quickly using criteria which are more flexible than those available on existing OPACs. These new systems will work more effectively if all the linking mechanisms are in place to provide a wide range of display options. The MARC format and AACR2 were designed to handle a situation which was complex and rigid and will need to change if they are to play their part handling information retrieval in what is in effect an information revolution.
Here are the case studies which I hope will reinforce the points that I have already made. Anyone with access to the Internet can go into BOPAC2 and repeat the searches to confirm that what I have said is correct or else examine the retrievals in greater depth using the display facilities that are available through the BOPAC2 Java applet. In all cases I have used MELVYL as the target catalogue and examples taken from the Guidelines. Some of these seem to be searches made on MELVYL but it seems that others are either simulated or taken from other OPACs.
Most of the examples that I have quoted are searches done at home on a six year old PC using a 14.4 modem. Most people will probably have more powerful facilities than mine and will get faster results.
Example 1 Tristram Coffin p10. The 'author' search for Coffin, Tristram resulted in a retrieval of 95. The select author option broke down the retrieval as follows:-
Looking at each of these alternatives in turn and then examining the full records for each by scrolling took a few seconds.
Going into the MELVYL catalogue through the Internet, the same 95 records were retrieved but they could only be viewed twenty at a time and it was not possible to examine the six different Tristram Coffins without a new search. Using a Java applet means that movement from the initial retrieval to one of the select options and then to some or all of the items selected either in full or in the shortened version is almost instantaneous. At any stage it is possible to obtain the MARC version of any or all of the retrieval. Another approach would be to select all the initial retrieval and then press the button to view all. This gives an title arrangement which, even with as many as 95 retrievals, is quite easy to scroll down, select the ones that are of interest and then view the full record for them.
Example2. Cummings E. p11. The problem of large retrievals is likely to become more important as more and more groups of catalogues (clumps in the UK) are set up using Z39.50. Cummings, E retrieved just under four hundred. The time taken to retrieve them was about two minutes and the time taken to get the java applet up and running was about double that. With my slow modem this is probably the worst scenario. However once the java applet is loaded movement through the retrieval is very fast using a number of different options.
Example 3. International Series of Experimental Social Psychology p29. With only 28 retrievals the user would probably scroll through either the author or title arrangement. An arrangements by numbers could easily be provided but if the number is known then the original search would include it and the search would go straight to the required record. The user's preferred display would be a subject arrangement and select subject is one of the options we hope to provide when we obtain more funding for development. An idea of how this would work can be obtained by calling up the MARC display for the retrieval and either scrolling through looking at the 650 fields or by doing a FIND search for 650 which will highlight all the 650s in the retrieval.
Example 4. Angels flight p58. Using 'title contains' gets a retrieval of 30 and the initial display is as good as that given in the Guidelines. Using 'select' and 'show these' gives a very effective initial display and it is very quick to move on to the full record. Calling up the MARC record by pressing F2 and then scrolling through looking at the 650 fields gives an immediate picture of the subject breakdown of the retrieval together with some variations in cataloguing practice. For example both cable cars and street railroads was used.
Example 5. Smoking p58. The Guidelines give an author and title listing and suggest that title is bad and author is better but the example was for a small retrieval and subject equals gets 358 retrievals. Using author or title initial options the retrieval is better because it could be scrolled through fairly quickly. Try 'select all' and then 'look at these' which gives a reasonable display. However a selection by the 650 field would be a considerable improvement.
Example 6. Guyton p64. This is a fairly simple search. Using Guyton , A as 'author', the search retrieved 64 records. 'Select all' and then 'look at these' immediately gives a breakdown by title and it is a matter of seconds to scan down selecting items which contain human and then look at these gives the full record which is a better display than that given by the Guidelines. A simple author title search would have course got the same result rather quicker but the example above was chosen to show how effective BOPAC was using less information. The full record gives the link between the editions where there is a change of title.
Example 7. Heard Museum p70. This example may not address the point that is being made in the Guidelines but it nevertheless illustrates the navigational powers of BOPAC2. An 'author' search under Heard Museum retrieves sixty four records which can be scrolled through quickly highlighting the items of interest and the full records of these can then be looked at. A more specific search under 'author' = Heard Museum plus 'subject contains' textile fabrics got four records and 'author' = Heard Museum plus 'title contains' Guatemalian got two records. I appreciate that this is not quite the point that the Guidelines make.
Example 8. Borden Lizzie p73 This seems to be an attempt to solve a fairly complex problem. Using BOPAC2 this would be solved by. using author OR title contains Lizzie Borden and this gets a retrieval of 42. The select all and then look at these arranges the retrieval in a sensible title order.
Example 9. Stewart, Don p84. This seems a cumbersome way to get at the information that seems to be required. A search on BOPAC2 on Stewart, Don retrieves 41 records. Select author shows that there are 5 different Stewart, Don authors as follows:- Don (24), Donald Douglas (9), Donald, H (1), Donald K (5) and Donald R (1). Selecting the Stewart, Don gives a good initial display which can be scrolled through very quickly. The instruction on the Guideline example to scroll up or down in various ways should not be necessary in modern systems. In BOPAC2 the chosen selection is scrolled whatever the size. Even a very large retrieval can be scrolled very quickly. [I have scrolled through a retrieval of over six hundred in a matter of minutes. This was to see how long it would take. In practice one of the select options would be chosen to cut down the retrieval.]
Example 10.crabs p91.This is probably a simulation as there are many more hits in MELVYL under any of the terms suggested. It is a useful and interesting idea which is also achieved by searching under the classification. Are there not a number of systems that already provide this? However, it has its limitations because in spite of authority control information in the 650 fields is not always consistent. This can be seem by retrieving 'crabs' either under 'subject contains' or 'subject equals'. This gives a retrieval of over ninety. By selecting the MARC format display and then FIND 650 the retrieval can be quickly scrolled through looking at the highlighted 650 fields to show variations in the subject indexing.
Much of the Guidelines is devoted to the display of subject information which is constrained by what current systems will provide and this is in its turn is based on the interpretation of existing standards. It assumes that authority control is in place and works effectively. This is not always the caseThe search strategy of the user is not always covered by the options provided by the system This means that what the user would like or what he can use effectively is often not available. Cataloguers themselves .have to work within the options provided by the vendors which in turn are governed by interpretations of existing standards. This results in searching options which are pre-coordinated. What the users would prefer is searching options which are post-coordinated and over which they have some control. After all the preferred heading to the user is the one that he is using.
With present systems the downloading of large retrievals is not usually possible and where it is the user finds little to help in navigating a way through it to find what is wanted. This will change for reasons that have already been given. The important point about this development is that the emphasis moves from the large database represented by the library catalogue to the retrieval which can be handled much more effectively at the local level.
OPAC display is extremely important and an input from both readers and systems staff is vitally important. Presumably this is being obtained.
In the hope that some of you may be interested in how we visualise the development of BOPAC2 the following is a short summary of the lines on which we will approach the funding bodies for the finance to continue our research.
There is a clear division between research and development in what we want to do. The research is mainly concerned with various aspects of information retrieval which have never before been applied to library catalogues. Since search engines and the relevance techniques are widely used we will be investigating how they can be included in the catalogues search options. We will also be investigating the linking of Internet resources to the library catalogue and integrating them with the catalogue users search strategy. Many libraries are beginning to include Internet resources in their catalogues but there is little progress in integrating Internet links into the catalogues search mechanisms.
The development aspects are concerned with implementing new features which we know will improve performance. These will include investigating alternative gateways to the one we are using, a more up to date version of Z39.50 providing servers themselves are supporting it or can be persuaded to do so and the rejigging of some of the Java program. For example to allow the production of retrievals which could then be formatted as bibliographies in a number of styles and perhaps most important of all provide additional sets of search options. BOPAC2 already works well but we would hope to improve it considerably from the development program and at the same time incorporate and test our findings from the research program.
This linking of two different strands is unique and is possible because the present version is already available on the Internet and is being used everyday all over the world. With development support we could move BOPAC2 from being a research project to an established service that would allow researchers unified access to an unparalleled collection of resources