The report begins by outlining the requests which are most often made to library catalogues and national bibliographies. Thus the report is based on these and aimed towards them, so that the practical aspects may be evaluated, whereas other more abstract theoretical formulations necessarily recall rigorous, logical coherence, and, however, do not exceed the limits of the indemonstrable, as these are their initial assumptions. Hence no pretensions to absoluteness, but a simple desire to respond to real situations and requirements. In this context the analysis of the relative value of the individual attributes in relation to the four functions requested from catalogues and bibliographies is appreciable

The global approach is of certain value; it takes the requirements of description, indexation, both semiotic and semantic, and of management into consideration; it anticipates search and use of records in different environments and contexts: catalogues, bibliographies, lists, in libraries and databases, in documentation and in retailing.

The choice of the entity-relationship model should be greeted positively for the clarity with which it permits complex situations to be analysed and the significant elements to be reported on in flexible and multiple use archives, for its proven validity in various fields and for its relative simplicity of comprehension and application.

The report is carefully drawn up: it rightly proposes to examine all types of documentary material, discovering and respecting the characteristics of each one, paying attention to innovations, above all technological, and recognising that continuous evolution always requires appropriate answers.

In the document some limitations are explicitly recognised, regarding problems to be faced on another occasion or subsequently (authority control archives, 'seriality', electronic resources, record format and filing systems) and it is important to point out that on all these issues IFLA is already active.

Open problems

Some wide-ranging problems on which to reflect in search of greater clarity are highlighted.

1. Entity of the first group: work, expression, manifestation, item
The insertion of expression to separate the two entities up to now mainly considered (work and publication, now called more generally manifestation) is particularly appropriate. The definitions are clear and the boundaries between the different entities fairly well-defined, but entities which are really different, and which it would perhaps be appropriate and not overly meticulous to distinguish, appear to have been levelled off under expression.
The various and different expressions of a work, if accumulated without distinction, do not permit a clear representation of their relationships.
In general there is an initial realization of the work (the first manuscript of a textual work, the first draft of a score, the first musical performance or staging of an improvised work...), a sort of original expression, which can be recognised as having the rank of initiator, which interests us here not so much to promote its specific characteristics as original data for comparison, as to determine the original type of work we are dealing with according to its nature (text, music, image, sound....). This first expression does not always leave a documentary trace (ancient works, test and music from oral tradition....) but it is possible to determine what its original type was. A poem created and told according to oral tradition does not have an original expression until its realization is proposed via a medium, normally written, or a sound recording: to this expression, to be considered the original, the recording of a letter or the transcription of the dictation may follow; music created on the instrument does not have an original expression until it is presented in musical notation or, if this phase is skipped, on a sound medium, such as the live recording of an improvised work.
The other expressions of that work (in addition to the original expression) may be distinguished in three types.
The first includes all those which realise the work by expressing it in specific forms which are intellectually different, but always using the same medium as the original expression (redactions, editions, performances etc. - whether subsequent or contemporary).
The second includes those expressions which, while conserving the same medium, change the canon or code of reference through a transposition (translations, musical transcriptions with different instrumentation.....).
The third includes those expressions which differ from the original expression because they use a different medium (performance of a written text, interpretation of a score, staging of a performance...).
Expressions of the first type are close to the traditional concept of new edition (varied, revised, corrected, enlarged...).
Expressions of the third type belong mainly the multimedial environment, due to works which lend themselves to different forms of representation.
Expressions of the second and third types are really expressions of other expressions (respectively of the first type and of the first or second type) of that work.
Taking the example of the Goldberg Variations (1), one may recognise the following structure according to the report: the work (abstract entity) is the variations as conceived by Johann Sebastian Bach, the expression is the form of musical notation which represents it in the score, the manifestations are the manuscript and the printed editions, the items are the single examples. Interpretations, such as that of Glenn Gould and all other interpretations (2) are similarly considered expressions, even though in another form, while the recorded editions are manifestations and the individual CDs, cassettes etc are items.
In reality each interpretation (expression of the third type) is not linked directly to the work, but is linked to it through the readable expression in the form of the score (manuscript or subsequent editions), if not actually through its printed manifestations, or a precise item of one of these (as is sometimes pointed out in the booklet distributed with the CD).
A transcription or arrangement for other instruments (with different score) is an expression (of the second type) and the interpretation of the transcription is also an expression (of the third type): the report, considering them all indiscriminately as expressions in the general sense, leaves the two intermediate stages of the passage as implicit: work - score - arrangement - interpretation, jumping directly from work to interpretation.
The different types of relationships foreseen in the report do allow us to distinguish and partly reconstruct the different roles of the entities (for example the relationship between transcription and arrangement is rightly foreseen as the relationship between expressions, not between expression and work, see p. 71), but entities of a different nature are associated indiscriminately among the expressions.
In a similar manner for texts, subsequent editions of the text (for example, modified by revision ...) and its translations - which more correctly are translations of a precise version of the text, even if this is not always easy or possible to verify - are without distinction expressions of the same work.
It is noted that more frequently the relationship indicated is between expression and work (see p. 74), due to the difficulty in ascertaining which other expression the one under consideration should be related to (on which score is the interpretation based?). The levelling out of the expressions seems therefore to be justified by the need to join the expressions whose intermediary is not known to the work.
This is a genuine difficulty, because in many cases there are not explicit declarations, but it does not seem adequate to also justify the complete elimination of relationships which are easily ascertained and of certain importance for much research. Again in the musical context: it is not indifferent to find the scores of the Goldberg Variations separate from interpretations (and within these harpsichord renditions separate from piano renditions), or performances of Boris Gudunov by Mussorgskij based on the 1869 version as distinct from those based on the 1872 revision, as well as from the version revised by Rimskij-Korsakov, and likewise those with the text in Russian as distinct from those with the translation...
Would it not be opportune, and possible, to correctly reconstruct all the relationships, starting from the expressions which realise the work according to the original medium (those of the first type, such as musical notation to stay with our examples) to which expressions of the second and/or third type (transcriptions etc. or interpretations, or transcriptions etc. and their interpretations) are correlated?
Or else would it not be appropriate, possible and perhaps more practical, to distinguish first of all expressive forms assumed by the work and within the single forms to distinguish the different expressions? To the musical creation (work), the score and the interpretation (notational and sound expressive form) would be related, to which would be related respectively transcriptions and arrangements (expressions of the first form) and the various performances (expressions of the second form); the manifestations (manuscripts and printed editions of the score, various editions of sound recordings...) would be correctly related to the respective expressions and thus it would follow that the items would be related to the respective manifestations. The work would thus be linked directly to the different expressive forms assumed, through these to the single expressions of each medium.
This type of hierarchical structure is not proposed in order to answer a requirement for classification, which here is perhaps little relevant (the entity-relationship model does not anticipate a dendritic graph, but rather a grid). It is indicated because, whereas for the descriptive function (which answers requests of type 2 and 3: identify and select) it is perhaps sufficient to declare and specify the attributes and relationships between entities in the record, in order to satisfy the indexing function, which concerns retrieval and ordering, an ordered distinction and aggregation of homogeneous entities is necessary (in forms to be decided: to give an example which does not require much imagination as compared to current cataloguing, with the indication in the heading of multiple related and relatable elements, such as: author, uniform title, form of document).
To put this another way: to the eye of those seeking an individual recording the correct indication of the relationship manifestation-expression-work is sufficient, even while obscuring some intermediate stages as implicit; but to the eye of those consulting an archive of recordings the ordered clarification of all the intermediate relationships is necessary for a correct search.
In the same way, for entities of the second group (persons and corporate bodies), the relationships are correctly distributed towards the entities of the first group (see 3.1.2 and fig. 3.2), thus linking only the entity responsible for the production, diffusion and distribution to the manifestation, whereas the creative function should be carried with the work, and that of intellectual modification and/or execution or interpretation should be carried with the expression, and only running through the relationships between the entities of the first group are these functions related to the manifestation and to the item: what is predicated by an entity is implicitly predicated for entities deriving from it, but not as in a genus-species relationship: I listen to Bach's music, Gould's sound, but Bach and Gould do not have a direct relationship to the disc, but rather indirectly as author and performer of the variations recorded by CBS and subsequently published by Sony.

2. Authority control

These reflections tend to clarify, but at the same time complicate the usual cataloguing systems more than the report itself does so already, while introducing the segmented analysis of entities of bibliographical interest.
From the clarification concerning the entities it follows that one must undertake authority control not only of the works (which is already additional to authority control of the headings only, to which we are accustomed), but also of the expressions, and thus we open the door to a new stemma structure which uses the literary histories of the subjects and bibliographical sources for precise reconnaissance of the ancestral and dependent items, including and in addition to the complicated and normally unexplained relationships between entities related to electronic documents.
Every time that a manifestation does not appear to be the only one existing of a work, it is necessary to go back to the work, highlighting the expression (or expressions) through which the relationship passes, as well as the relationships with other expressions, works and manifestations. Thus an increasingly complete mapping of the works, their expressions and manifestations and of the relationships which link them together and to other entities (persons and corporate bodies first of all) takes shape.
To this should be added the fact that some works are part of another work. The Goldberg Variations, for example, are part of the Clavier-Ubung, whereas in their turn they are made up of an aria and thirty variations: each of these entities, which may present itself as an expression or manifestation isolated from its context, should be considered a work, and they are all connected by relationships of the type whole/part. The mapping which was referred to must therefore also provide for the existence of parts of works (dependent or independent, as the report specifies) not only in the obvious sense of anticipating explanation of the appropriate relationships in the records but also in the sense of reconstructing the stemmata.
This mapping is at the same time the origin and the result of bibliographical work: to have precise stemmata of the works of authors and of their expressions already ready available is a convenient premise for the compilation of bibliographical archives, but it is only through this work and the study of the tradition of the texts that these 'family trees' or stemmata are created, completed and then updated. And nobody has the 'philological' competence (or the tools to undertake this) in every field of knowledge, in every documentary context, whereas here it appears that those compiling bibliographic records must be able to base their work on reference works which in reality only exists for the most consolidated traditions (for example in the case of classical music with the catalogues of composers), and, when they exist they are not easily available; or else cataloguers must improvise investigations, comparison and collations of and between works, expressions and manifestations and their parts, straying into the field of the scholar of the subject.
This is not a new problem, but rather a new version of the cataloguer's dilemma, normally resolved in library catalogues with the criteria of economy and evidence of publication, thus making do with the most easily available references to original titles or to titles of previous editions, often checked with scarce enthusiasm.
The new model, on the other hand, proposes methods whose clarity requires the maximum precision and completeness, if the results are to effectively correspond with the conceptual rigour of the formulation; but moving freely over contexts which continue to widen, this precision and completeness can only be the goal of long and careful work shared by individuals who are able to cover partial sectors and refer to all the instruments available with specific competence.
The means of establishing the results of this work of recognition of the works and their expressions cannot be other than a form of authority control, through which these entities are named univocally (uniform title) and correlated with reference to other forms (equivalent titles, correlated titles...)
Thus descriptive elements whose transcription is traditionally (also according to the report) left to the free form of the publication, must be formalised univocally in order to be recognisable as equivalent. What already takes place in cataloguing for the names of the authors, transcribed as they appear in the document for the recognition of the publication, but rewritten in a uniform manner for recognition of works by the author, must also take place for all the other elements to which in addition to the representative function of the object described it is desired to add the function of information retrieval: titles of works, of parts of works, of series, names of publishers, of locations..., and furthermore titles of expressions... At the same time it will be necessary to create links with as many archives as there are categories of entities interested in this aspect, in which the equivalent forms are connected with each other.

Furthermore, in the report the case of manifestations which embody several expressions of several works is foreseen. To stay with the musical examples, in which various pieces are normally performed, a piano recital or recital by an opera singer, in the recording on disc which embodies it figures as the manifestation of several expressions (given that the performance of each work is considered as a separate expression - rather than considering the recital a single expression), each of which is the realization of a different work (which may also be a single part of a work).
At this point it is clear (and this is not the most difficult example, it is enough to think of the complexity of multimedial documents, to the parts called 'systemic' in the report) that the intertwining of relationships becomes more and more complex, multidimensional, with ramifications of the works to the multitude of their manifestations, but also of single manifestations to the multiplicity of works contained in them, and hence difficult to trace back to simple patterns.

3. Economy

In the report the principle of economy is implicitly present, when the writers consider the reduction of costs for bibliographic records and when they leave it to the discretion of the bibliographic agencies whether to pass over certain data not considered essential. Nevertheless the report does not propose a saving of work, but rather an increase; breaking down into entities, research into relationships and mapping, formalization and authority control. Is this an apparent contradiction because it is foreseen that subsequently the increase in work will be compensated by greater wealth, better precision and hence improved practicality of the archives? Or is the model only an ideal conceptual point of reference, which can guide the more complicated cases of cataloguing which however remains fundamentally traditional? The abbreviated solutions permitted where the use of certain elements is not clear would appear to confirm this interpretation. But if these shortcuts become too widespread, as is it easy to predict may happen, will this not defeat the overall objective? And is it possible for records with exhaustive explanations of entities and relationships to coexist alongside simplified records without giving rise to confusion and misunderstandings?

4. A fifth task

Apart from these considerations concerning the viability of the model, one again asks oneself: if bibliographic records are to represent the complexity above and catalogues are to permit the correct search of each entity, it would appear logical and important to explicitly add as a fifth task that the catalogue should permit, as well as finding, identifying, selecting and obtaining, that of surpassing (going beyond), of correlating the entities found (identified, selected, obtained) in the sense of permitting, indeed of facilitating the passage of data of a first unsatisfactory, or not fully satisfactory search (or else, on the contrary stimulating) to other records of data related to the first. This is a function which gives a sense of the organic unity of catalogues and bibliographies.
In the model it is essentially the role of the relationships to carry out this function, which is effectively cited, but without particular emphasis, with the by now commonplace metaphor of 'navigating' in 5.1 (p. 56).

5. Relationships

The identification and listing of possible types of relationship appears both analytical and precise, with even the distinction between dependent and independent parts, between referential and autonomous expressions. Their essential role is to reveal, specify and designate relationships between the entities; in this way they facilitate their distinction and identification: expressions are identified in relation to the works and the manifestations in relation to the expressions...
Once this is recognised one asks oneself: how are the relations represented?
They are found within the same record expressed linguistically (a typical example are the notes on titles translated), or else with the association of descriptive data (the name of the publisher within a description, the name of the author connected in the heading...), or also unexpressed but implicit (when the title of the manifestation is at the same time title of the expression and/or of the work). Here, the more explicit they are, the easier is the correct comprehension of the object described in its literary and actual context.
But within the record should all the relationships be explained? Should they be made operative, that is should the two entities of the relationship be named in an identifiable manner (see p.57), to avoid incapacity of demonstrating search routes from one entity to another? And in this way how greatly would records and archives swell?
The relationships should be the link which joins, the bridge which connects different entities, hence linked together but external as regards the entities.
In this case the question arises: where are the entities to go? and what is the relationship between entities and records?
A complete representation of the entities and relationships appears possible only by creating as many records as there are entities considered, supplying them with the attributes relative to the entity itself and links with other related entities; thus each entity appears only once in the archive, but its record may be cited by any other entity in relation to it.
Instead it appears that the logical exercise of distinguishing works, expressions, manifestations and items and the entities of groups two and three (authors and subjects) is swallowed up in a record still of the traditional type, which associates data relevant to other entities around those of the manifestation, without however presenting their complete attributes, or permitting the autonomy which would allow them to be the branch point for relationships with other entities.
Indeed, if the expression is not recorded by itself, but scattered around the records of its manifestations, it may not fulfil its function of bringing together all the manifestations which embody it, nor may the work scattered around the records of its expressions carry out its function of bringing together all the expressions which realise it.

With this aim, among other things it will always be necessary to apply two-way relationships, not only to trace back from the item to the manifestation, expression and work, but also in the opposite direction in order to reach all the expressions of the work and from each of these all the manifestations and items.

Hence it appears the record must assume a new format which moves away from both the unified sequence of the eight areas foreseen by the ISBDs and from the card with description completed by heading, callnumber and tracing, as well as from the traditional unified forms of bibliographic citations. Instead it would be like a constellation made up of a central nucleus to which other entities are connected in various ways, whose satellite role may be transformed, by moving the focus, into the central nucleus of another constellation. Both the necessary and convenient dimensions of the constellation to present in single displays and in the different bibliographic records extracted from this universe should be examined (the nucleus with how many and which links? - the weight given to the attributes and the relationships in chapter 6, and the basic requisites for national biographies in chapter 7 are indicative of this). The catalogue would no longer appear as a list, but as the universe of this grid, that can be travelled over via adjacent stops starting from any point; a bibliographical list would be a choice of route across the grid; a single citation a constellation more or less articulated.
The need to foresee different levels of completeness of the records is again evident, according to the context they are designed for (for example, a citation as compared to library catalogue) rather than primarily according to the availability of the compiler. Indeed, a simplified bibliographic record abbreviated to minimum level may not disregard the analysis carried out on the entities involved in the process of undertaking the physical units to record: the examination of the item as an example of the manifestation always refers to the level of the expression and the work to identify intellectual responsibilities, the relevant subjects. Furthermore, simplification must never cancel the link, the indication of connection with other correlated entities.
This is a central issue, still to be examined more closely and explained.

The consideration made above concerning the economy and hence the effective viability of the model may here be raised again and confirmed.

Observations on more restricted aspects and problems

A last series of observations concerns more particular aspects, single points of the report which appeared to be unclear or little coherent but which do not in any way invalidate the value of the document.

6. Entities of group two: persons and corporate bodies.
The issue is developed little, because the parallel use of authority files is supposed.
It would be possible to highlight at least the possible relationships between entities of the group: between person and corporate body in the case of affiliation or representation ("is a member of, is a representative of...", "has as its member, has as its representative..."); between organisation and organisation in the event of subordinated body or section ("is a body of, is a section of ....", "has as its body, has as its section ....") and in the case of succession ("has as its successor...", "follows ....").
In 3.2.5 and 3.2.6 the persons and corporate bodies are related only to works, as opposed to 5.2.2 and figure 3.2 where they are correctly related also to expressions, manifestations and items.

7. Entities of group three: subjects.

From the definitions, explanations and examples it appears that concepts are only abstract concepts (single nouns), objects, events and places are individual (formulations or proper nouns, singular). Are concepts which represent a class of objects, events or places (for example, ships, earthquakes, deserts...) not foreseen? In 4.9.1 form is used which is not coherent with that used for events and places: "a building, a ship", probably due to misprint.

8. Aggregate and component entities

Paragraph 3.3 widens the meaning of work to aggregate entities and component entities, to be related in whole/part relationship. This is correct conceptually, but it would perhaps be appropriate to take the opportunity of such an in depth and rigorous analysis to clarify and better limit the possibility of considering such entities as works, offering a contribution to a dilemma which has not been resolved. Can a publisher's series be considered a work or by its nature does it not belong rather to the competence of manifestation? For various motives, it is traditionally considered convenient to use the series as a group of monographs to be kept unified (for control of purchasing in the library, for the arrangement of shelves in the bookshop...). In reality belonging to a series mostly means the sharing of a series of characteristics common to books of the same publisher: size and binding, graphics and layout, a certain cultural approach, depth of examination, sometimes subject matter and obviously the title of the series. In order to belong to the competency of work or expression on the other hand, the series should qualify essentially as a creation or intellectual or artistic realization, elements which are undoubtedly present in the work of directing or drawing up of the series, but which are always subordinated to the publication of the monographic works which make up the series. On the other hand, a single work may be part of a work-series if it is created and dies within it, given that the link between works is stable, namely if the single work is part of the work-series it will always be a part of it, in whatever expression and manifestation.
The current tendency to consider the title in the series as autonomous, with linking monographic titles, is an operation designed to gather together all the titles which appear in the series, but looks more like finding all the titles published by a publisher. As an entity the series has more to do with the concept of editorial initiative than with the concept of the work. Where the titles in the series are not distinct from the titles of monographies or of periodical publications, serious problems of identification and distinction of bibliographical objects would arise.
There is also the problem of defining what a work is when there are whole/part relationships. Indeed, in these cases if both the whole and the part are works, linked by a simple relationship which does not even establish which the main entity is, it does not appear possible to organise the catalogue correctly. For example, if Du coté de chez Swann and Il giardino dei Finzi-Contini may appear indifferently as works in themselves and as parts, respectively, of À la recherche du temps perdu and of the novel Romanzo di Ferrara, with a relationship without head or tail, it will be extremely difficult to find all the editions without making use of multiple searches.

9. The intellectual, artistic etc. responsibility as regards the entities of group one is resolved by establishing relationships with entities of group two. The formulations of responsibilities are then presented as attributes of the manifestations, p.31 and 4.4.2) as data exhibited by them, whereas the work is an abstraction and the expression does not have complementary textual features, which is the reason why these two entities do not have a particular presentation of indications of responsibility to transcribe as elements of description or recognition. Thus it is possible that names carefully distinguished as not relevant to a manifestation, but rather to the work or expression, nevertheless appear as attributes of the manifestation. The ISBDs, which supply the pattern for most of these attributes and which by their nature do not apply an entity-relationship model, but only distinguish between areas elements relevant to the different entities, here show that their function is limited to the description of the publication (i.e. manifestation). The model includes and incorporates them in a bibliographical record which, through the relationships, redistributes the elements in the appropriate entities and with univocal forms chosen according to the search. Without remixing the different functions, it nevertheless appears important that the ISBDs be renewed in order to better reflect the nature of the information they supply.

10. Due to the completeness of the obtain function and given the predictable application of the model to magnetic records which are constantly being updated, it would appear useful to add availability to the attributes of the item, namely the indication of temporary unavailability when the document is in use, as occurs in systems which integrate the catalogue with the management of the circulation of the documents.

11. Finally, classification numbers are cited in 2.2 and then only reconsidered sporadically (as organisational elements in national libraries on p.115 and as attributes of subjects on p.132-133) and the abstract (attribution of the expression in 4.3.9; why not of the work?). It would be interesting if their function and logical placement were clarified.