Creation of various semantic networks for semantic networks contain all statements-offall by the wayside. Representative questions department, in order not to let any information here. parts of total area of attention. These from the user-departments can also be noted fact and requirements issued by the user
Core of Model to-day business its priority, nor is so understandable therefore as incapable of eectiv ely contradicting the ’high-ying’ ideas of their DP-colleagues. The result terminology and a certain generalisation (see 1.1). Although the user departments are consulted during The user department sta usually see themselves the impression that the DP-sta’s ideal model is dethe analysis phase, in practice one is often left with is mostly a model which gives the impression of abthat the user-departments can work with it. the creation of the data model requires a change in solute perfection, but which neither makes the dayveloped. This trend is strengthened by the fact that all-too-seldom leads to recognition of this ’inadeno longer the topic of discussion (in which the quesmeans that the model, which really should be the tion of understanding certainly arises) - rather, one The relatively low expressiveness of an E/R-model basis for a common understanding, often becomes a problem of understanding. The real world is then quacy in meaning’. Often this inadequacy is compensated for by an overkill of interpretation, which
Developer of entities. Let’s leave this point for the moment methodology. In this way, entities such as ’Total technique. Very often, the interviewer’s question is the weakness in the developer’s understanding of the - it will be touched upon again in point 1.8). Anbehaviour is also to be seen in a completely dieren t instead of "Which process stati occur in practice nical background and mean tables or les instead other weakness is the missing experience in interview the resulting excellent representation often disguises formulated like "And how can I show that in E/R?" Turnover’ and ’Turnover per Customer’ can actuconcepts as entities, instead of taking the expressive ally be modelled. Most developers tend to model form, where the developers come from a very techThe possibilities of graphical development tools and character of entities in general into account. (This let’s leave E/R out of it for the moment?". derstand the ’fact-content’ behind them. discusses entities and relationships, whose meanings of interpretation and alteration when trying to unare comparatively trivial and thereby are a matter nise relationships between them that are held to be necessary, then the model seems to them to be commodels quickly become too detailed and dicult to cannot follow it through to the lowest detail. had to be documented. The eect of this is that the to build every case imaginable into the model, as if plete and correct, even though in many cases they cases comparatively simple to represent, Data ModSince the daily business of a company is in most the knowledge for treating each of these cases really understand - so much so that the user-departments, more or less make this judgement on the basis of who really should judge the model’s ’correctness’ ’gut-feeling’. If they see well-known terms and recogelling projects often rush headlong into attempting
to requirements, rather use their own ideas as basis for the ’ideal model’. The standardisation of terminology (mentioned untion’s success so much (This could certainly be the dictate the terminology of the remaining employees. larly aect the importance of DP for the organisasubject of heated discussion both in theory and in It implies however, that 0.5 - 2 % of the company can DP tends to over-estimate itself in many organisaan organisation’s leadership). This obviously false der point 1.1) which the DP-Department carries out plementation of standards and norms much more. lined in 1.3, means that DP doesn’t model according judgement of one’s own situation aects the imtions. This inaccurate estimation doesn’t particuduring data modelling is here an excellent example. entities with the same attributes are hastily made one, without considering that they express a classicured where performance considerations were incorporated into the E/R-Diagram. The problem, howcannot imagine any way to represent the characterground (e.g. Application Development), they nd cation on a logical lev el. edge when modelling. In the past, many cases ocSince most modellers come from a technical backistics of entities other than with attributes. Two ever, goes much deeper than that. Most modellers it extremely dicult to ignore this tec hnical knowla semantic network. It’s quite possible that other, ment’s) tendency to express himself in concepts. This means that in the initial Data modelling phase, to best eect the developer’s (and the user departone creates a model of these concepts in the form of terms with knowledge representation methodology. tic networks was the easier way for him to come to development of a Data Model (relational or E/R): the Data modelling world, moving towards semanThe approach this solution takes is basically to use more modern, representations are more suitable for The author makes the following suggestion for the this task. However, since the author has his roots in
Consolidation of the various networks. The aforementioned networks are consolidated. tation is not subject to the restrictions of the E/Rtioned problems. The interviewers interviewmodel. The developement of the model is also docThe suggested methodology solves the aformentechnique is positively aected, because his annodiscovered during the analysis phase, is held in the umented, whereby the supplementary information model.
concept The cabability to consolidate the various parts stays relatively small. means that the model in the user-department mantics of these associations are not interpretationtheir symbolic meaning, rather by a relatively formal context. This has the advantage that the sedependent. Note that these associations are not dened by be re-created and checked in reviews. The usermodel’s content, thereby avoiding ’ideal strucdepartment sta can concentrate more on the The selection process (what’s an entity?) can tures’. essary for the consolidation. The following two types standardized, unied represen tation of the networks works developed by several developers possible, a of associations are allowed to be represented by lines: This restriction forces the unied represen tation necsociations are allowed, represented by lines; all other To make the consolidation of several semantic netrelevant concepts and associations appear as nodes. is suggested. This means that only two types of asGeneration of an E/R-model. The user denetworks. This model is then the basis for the partment requirements can be generated using all of the semantic networks. The E/R-Model creation of the relational model. tested using the requirements represented in the can be worked on using this basis and can be
There are, however, also disadvantages. Type 2, which denes the in tention. ticial terms’ is restricted - the terminology can The tendency to strong generalisation and ’arstill be understood by the user department. concepts, and not other elements such as entities, in order to express the facts in the required manner. what less formal representation is chosen, which is the interviewer requires considerable concentration in its entirety. Furthermore, during the interviews, If one uses a strictly formal representation, as sugeven if a less formal notation is used. subsequently translated into a formal model. In practice, however, during the interview a somegested above, the model becomes dicult to grasp Note that the principle elements of the model are guage necessary. Rather, the individual terms are cross-referenced to one another. This means that there is no unication of lan