Computer-Aided Globalization of Advertising By Expert Systems

ABSTRACT - Globalization of advertising is a multi-stage process requiring a very comprehensive knowledge base with data on cognitive and emotional behavior in the different target countries. Knowledge based computer programs as the CAAS are suitable to provide the needed data.

Citation:

Werner Kroeber-Riel (1993) ,"Computer-Aided Globalization of Advertising By Expert Systems", in E - European Advances in Consumer Research Volume 1, eds. W. Fred Van Raaij and Gary J. Bamossy, Provo, UT : Association for Consumer Research, Pages: 110-117.

European Advances in Consumer Research Volume 1, 1993      Pages 110-117

COMPUTER-AIDED GLOBALIZATION OF ADVERTISING BY EXPERT SYSTEMS

Werner Kroeber-Riel, University of the Saarland, Saarbrncken, Germany

ABSTRACT -

Globalization of advertising is a multi-stage process requiring a very comprehensive knowledge base with data on cognitive and emotional behavior in the different target countries. Knowledge based computer programs as the CAAS are suitable to provide the needed data.

1. Globalization of Advertising

Through the creation of a common market in the European Community, discussion on global advertising strategies receives new impulses. At many a university, European Institutes are founded dealing with marketing on the European markets in research and teaching, especially with the globalization of marketing.

We start with a model of multistage globalization of advertising (see Table 1). Essentially, this model presents two stages:

1. Stage: Globalization of the advertising message: Here the question poses whether or not the same central message may be used in different countries.

If advertising aims to position a product or a service, the central message refers to factual attributes as "economy of a car" or to emotional benefits as "prestige of a car". Whether or not the informative or emotional message can be globalized depends above all upon the cultural meaning of the offered products or services, especially, however, upon the fact in how far the central message conforms with the consumers' life styles.

It may be supposed that the central message may often be globalized within the European countries. However, there are also considerable cultural differences calling for an adaptive positioning of products and services, i.e. conveying differing factual attributes or emotional benefits through advertising.

Especially consumer's emotional behavior shows considerable differences in the European countries thus impeding the globalization of the emotional message.

Examples: According to a pilot content analysis by Granowski (1989), French advertising contains more than twice the erotical appeals (15 %) of English advertising (7 %). Or: Social happiness is often conveyed through German advertising (11 % of the emotional advertising) and only very rarely by English advertising (2 % of English advertising).

A fundamental cause of these differences are deviant "feeling rules" in the European countries. These cultural rules determine in which situation which feelings have to be perceived and expressed (Kemper, 1991, pp. 323 ff.). The choice of emotional appeals for the positioning and the verbal and pictorial expression of emotions has to consider these cultural rules.

2. Stage: Globalization of execution: The same central message may be executed in different countries verbally or pictorially in the same or in a differentiating way. For example, the message "brand X conveys feelings of freshness" may be expressed in various countries through the same pictures of swimming scenes. However: Executing "freshness" by pictures of ice or drizzle may induce positive feelings in southern countries and trigger negative impressions in northern countries.

In Europe, the global verbal execution is constricted by the insufficient proficiencies in languages across the borders. The English language as common nominator is but insufficiently known in the big European countriesBin France, Germany, Italy and Spain (TTlTperformance, 1988).

Therefore, it should be ascertained that language and pictures used to convey a message evoke sufficient similar effects for a globalization in the different countries. If this is not the case, the verbal and pictorial messages have to be harmonized with the culturally determined understanding in the target countries. In this case, advertising follows the motto "global heart and local faces".

2. Expert Systems for Advertising

At the Institute for Consumer and Behavioral Research of the University of the Saarland, a complex system to develop and evaluate advertising called CAAS (Computer Aided Advertising System) was developed comprising three main subsystems (Table 2):

- an expert system to evaluate advertising (diagnostic system)

- an expert system to develop advertising (search system)

- a computer system for picture and word processing (picture manipulation system).

The diagnostic system is a typical expert system to evaluate print and electronic advertising. The knowledge is represented by a production system. The rules representing advertising knowledge relate firstly to the effects of advertising strategy and secondly to the effects of verbal and pictorial execution. They were derived from the present literature on advertising, especially from the book "Strategies and Techniques of Advertising" (Kroeber-Riel, 1991). Uncertainty of the implemented knowledge is taken into account. In detail, knowledge engineering, implemented rule system and inference components of the system are documented in current publications (Kroeber-Riel and Esch, 1992, Esch und Kroeber-Riel, 1992, Esch, 1990, Lorson 1992).

The rules of this expert system are based on a hierarchical model of advertising effectiveness structuring the advertising effectsBdepending upon the advertising goals and upon the involvement of the target groupBin "modules of effectiveness" more or less important for advertising success. For the aggregation of the ascertained effects the system differentiates between necessary effects (non-compensatory aggregation) and sufficient effects (compensatory aggregation).

As a tool, GoldWorks is being used, as hardware a 386 PC with 16 megabyte working storage expansion.

Special care is given to the surface of the system. Two screens are available to the user: one screen to be used for the interaction with the computer (and) to feed data into the computer, and a second one supplying the user with definitions, explanations and examples for a better understanding and answering the questions for the evaluation of the advertising material.

TABLE 1

MULTI-STAGE GLOBALIZATION OF ADVERTISING

TABLE 2

COMPONENTS OF THE CAAS (COMPUTER AIDED ADVERTISING SYSTEM)

The development of the system is completed. Further information will be presented in this volume by Esch (Section A6). Currently the diagnostic system is practically used by those firms having supported the development of the system (as BASF, Schering, Procter & Gamble, Jacobs Suchard, Underberg). The further updating and commercialization of the system have been taken over by Allcomm Advertising AG, Basel, a subsidiary of Ciba-Geigy.

The search system serves to support the user in searching pictures for advertising, expecially new and effective pictures for the positioning of a brand by emotional benefits as "natural", "rural" or "fresh".

Hence, this computer system does not have an analytical objective as the diagnostic system but a heuristical, "creative" objective. Consequently, the knowledge base and the surface of the system are differently constructed than those of the diagnostic system: Above all, the knowledge is represented in form of semantic (associative) networks as well as in form of heuristical search rules.

A 386 PC with 16 megabyte working storage expansion is being used as hardware, as softwareBwith user surface MS-Windows 3.0BKnowledgePro, Version 2.0.

The philosophy of the system is to stimulate the user's own creativity and to prevent that only few and exchangeable pictures for advertising are being developed. Thus, the system fulfills two demands:

(1.) Stimulate the user, show him ways for the creative search process, but don't offer him/her ready-made solutions!

(2.) Provide a multitude of visual ideas and clues for the creative search process!

This heuristic procedure becomes clear by looking at the first step of the search process. The user begins to search pictorial ideas for a positioning concept (search concept) like "natural product". Initially, the user is aksed by the computer to utter pictorial ideas and to enter them into the computer; only then the computer delivers further visual ideas.

In the course of this interactive brainstorming the user receivesBin addition to his own ideasBover and over again new visual ideas procured by the kowledge base of the search system and stimulating his/her creative search process. His/her own associations are being storedBas far as they are not yet available in the computerBfor consequent use. Thus, they supplement the knowledge base.

To stimulate the user, more than 70.000 associations connected with 220 central positioning concepts (as biological, Caribbean, comfortable, exclusive, international, modern, natural, reliable, tender, etc.) are stored in the knowledge base all clustered in empirically measured German association norms (Petri, 1992). The clusters include abstracts as well as concrete visual associations and color associations. (Therefore, the stored associations are considerably higher structured than those in the search system "Idea Fisher" (compare Watson, 1989)).

Of course, the knowledge base cannot include associations for all possible search concepts. If the user searches visual ideas for a concept not stored, the computer suggests semantically adjacent search concepts for which visual associations are available. These adjacent concepts are found by means of a dictionary for synonyms and by calculating the semantic difference between the search concept of the user and the available concept in the computer system.

This procedure only constitutes the first part of the search process. Next the computer offers creative techniques and aids to continue the search of ideas. From a wide ranged menue, the creativity technique "improving with colors" may be chosen: Upon calling this program part, the colors to be associated with the search concept are shown on the screen. For example, the concept "natural" associates with the colors "green" (56 %), blue (27 %) and brown (22 %). The user may then search for pictures in which such colors dominate. Such pictures will often procure a natural impression.

In addition to this program part,

1. creative search for visual ideas

the entire search system comprises three additional parts (modules):

2. search for persuasive motifs of pictures

3. search for pictorial schema attributes

4. search for effective pictorial execution.

Imagine again the search for visuals to express the positioning concept "natural". The output of the second module may be a picture of a meadow with a fairy in the foreground instead of the idea to visualize "natural" only by a meadow as found by the first module: A fairy has a strong behavioral impact appealing to an archetype.

The third module makes it possible to find visual attributes for this picture ensuring that the picture of meadow and fairy is really understood in the sense of the search concept. For example to enhance the impression of "natural", flowers and butterflies are integrated in the meadow, the fairy receives a blue dress.

The fourth module serves to develop an execution of the picture adapted to advertising purposes. For example, to improve the memory effect, the interaction of the pictorial elements is intensified: The fairy picks flowers, etc.

Within the next few months, the search system will be completed and will be delivered to those firms having paid their subscription price.

Pilot tests showed that the creative search process by means of this system is considerably accelerated and procures more solutions of a wide variety. Based on the foundings of creativity research "quantity produces quality" chances increase to find innovative and effective pictures for advertising.

The pictorial manipulation system is being used to execute pictorial ideas and to manipulate pictorial material.

All subsystems may be used interactively. So a completed ad or an animatic can first be evaluated and then new pictures in exchange for a weak pictorial motif or a weak scene may be developed with the assistance of the search system. OrBstarting with a positioning conceptBan advertisement is developed with the search system and the picture manipulation system which is then "tested" with the diagnostic system.

3. Globalization of Advertising Through Expert Systems

Whether and how far advertising can be globalized may only be decided after obtaining insight into the cultural similarities and differences of the target countries.

With regard to Computer Aided Advertising System (CAAS)Band generally with respect to corresponding diagnostic and search systemsBthe following questions have to be answered as a matter of priority:

1. Is it possible to use a diagnostic system to evaluate advertising of other coutnriesBeither by users in our country or by users in other countries? Which modifications are necessary to carry out such evaluations? Neibecker discusses these problems in the following contribution.

2. Which are the requirements the diagnostic system has to fulfill to judge whether and how far an advertising campaign is suitable for global use?

Corresponding questions have to be posed concerning the search system: In which way, especially with which knowledge base, can advertising be developed for foreign countries or globalized by means of the search system?

Apart from the use of expert systems and reduced to a common denominator, the key question is: What kind of knowledge is necessary to develop and to evaluate advertising to be used in different countries?

To practically answer this question, empirically established information on cultural distance to the target countries is needed. Above all, this may be acquired through direct measurement of

- life style

- verbal behavior

in the different countries and target groups (indirectly also measurable through content analyses).

Numerous studies on these subjects have been conducted meanwhile. For example, in Paris, the International Research Institute on Social Change (RISC) identifies life styles in European countries to inform marketing on the more or less global responsiveness of consumers (Woesler de Panafieu, 1988).

Cultural values are likewise recorded crossculturally and compared. They may be understood as determinants of lifestyle an example of which is set by the comparing studies on materialism by Ger und Belk (1990).

We here consider verbal behavior. A preferred procedure for recording are free association experiments: Subjects are asked to spontaneously respond to a given stimulus word as "natural". Example: To the stimulus word "natural", German students react at first with the following associations (percentage of associations related to all students):

TABLE

Instead of the frequency of all associations, very often the frequency of primary responses is indicated; these are associations uttered as first word upon the stimulus word.

Depending upon the goal of the procedure, subjects can also be instructed as to how they should associate. Since we were, above all, interested in visual associations for our CAAS search system, we consequently gave the subjects an imagery instruction: Subjects should present such visual associations coming to mind in connection with the stimulus word.

Based on this data basis, various indicators may be found informing on the cognitive and emotional behavioral patterns in the different countries, which can be used to answer globalization questions.

Association norms: These are association patterns distributed among a larger target groupBfor example among students or workers of a countryBi.e. associations appearing with a certain frequency (certainty) and consequently reflecting cultural characteristics.

Association norms may be understood as more or less standardized cognitive structures often called associative networks by cognitive science and being used to represent knowledge in the memory or in the computer (Stillings, Feinstein et al., 1991, S. 26 ff).

They are applied in consumer research to explain mental representation of products and the ensuing purchasing behavior (Grunert, 1990). They also give insights into the comprehension of positioning concepts as well as of central messages of advertising. For example, the network around the concept "Italian" (Table 3) illustrates a perceived national stereotype and reveals considerable cognitive and emotional responses to "Italian".

Many crosscultural studies are based on association norms. Well known is the collection of association norms of American, French and German students edited by Postman and Keppel (1970) and the work of Scalay and Deese (1978) on "Subjective Meaning and Culture" including comprehensive methodological remarks.

Of special interest are the derived findings on the culturally perceived characteristics of objects and the pertaining emotional relationships. Table 4 shows the meaning of "bread" of Californian, French and German students: In California, the mental representation of bread was more homogeneous than in the European countries indicated by few dominant associations. Moreover: "Bread" was not associated with "hunger" as in France or Germany. In California, "butter" was the strongest most distributed association to breadBbut does not appear in the association norms of the French students.

All in all: Association norms and indicators derived from these association norms are operational specifications for cultural patterns of behavior, which can be stored in the knowledge base of expert systems and may be used as follows:

Using the diagnostic system, it would be possible to find outBon the basis of the stored association normsBto which degree the central message fits the cultural understanding, whether or not the cultural understanding includes unintended or misleading message elements. Already association norms for a few hundred keywords would suffice to procure an effective assessment of advertising messages.

Association norms of the search system would enable the user to develop advertising starting with the cultural meaning context to a search concept in target countries. Metaphorically spoken: The user is given the possibility to poach in the associative networks of the people in the target countries to receive suggestions for visual ideas suitable for advertising in these countries.

A more flexible and efficient usage of association norms will be reached on a higher level of data aggregation: The stronger or weaker similarity of association norms and hence the mental distance of the key concepts can be expressed by overlap coefficients (OC).

One version of this overlap coefficient can be calculated as follows

EQUATION

i.e.:

i and j = the association distributions to be compared

c = the common elements appearing in these distributions

c_l = the lowest relative frequency of one of the common elements c

TABLE 3

ASSOCIATIVE NETWORK FOR "ITALIAN"

TABLE 4

CROSS-CULTURAL COMPARISON OF ASSOCIATION NORMS FOR "BREAD"

TABLE 5

ASSOCIATION NORMS OF GERMAN AND FRENCH STUDENTS: HIGH OVERLAP COEFFICIENTS FOR "FRESH"

TABLE 6

ASSOCIATION NORMS OF GERMAN AND FRENCH STUDENTS: LOW OVERLAP COEFFICIENT FOR "TRANQUIL"

The overlap coefficient for the German and French associations norms is high as to Table 5, however low as to Table 6.

The stimulus word "fresh" (frisch, frais) induces very similar associations. Among the eight first ranked associations, the same associationsBas "fruit"Bappear five times in the two association distributions.

The presented association norms for "tranquil" (tranquille, ruhig) contain only two common associations. First ranked German associations are "sleep, night, church, cemetary"Ball missing in the French associations.

Accordingly, low overlap coefficients show high differences in the meaning of the stimulus word in the different countries. This result may be taken advantage of both in the diagnostic system and in the search system:

In the diagnostic system, overlap coefficients can be stored relating to the meaning (the cultural understanding) of positioning concepts and key words of advertising of the different countries. They indicate to which extent a globalization of the message is possible (high overlap coefficients) or whether the globalization is difficult and needs further analyses (low overlap coefficients).

In the search system, overlap coefficients related to the stored search concepts (as "naturual") can be revealed and thus deliver clues whether the search for visual ideas for advertising in the different countries should be in the same or in another association space. High overlap coefficients make it possible to develop global verbal and visual ideas right from the beginning. Examples are the strongly overlapping images of "natural" or "fresh" in Germany and France allowing the same message execution in both countries.

A further group of indicators are the highly standardized associative nets manifesting in schemata and schema attributes and also in pictorial symbols.

If the visual primary associations to a stimulus word show high frequencies, people connect this word with rather the same memory images. Examples are the associations for "magic". 36% of the primary associations and 75 % of the total associations are manifested in the visual of "magician". If the continuous associations to "magician" are determined, the following main visual attributes can be ascertained: (1.) peaked magic hat, (2.) (black) cloak and (3.) requisites, especially "magic wand". Corresponding pictures, also single pictorial elements, convey a quick and comprehensible impression of "magic" (Maas, 1993).

The verbal data and picture bases of the the third module of the search system contain numerous schemata and visual schema attributes related to the search concepts. If the user wants to obtain a typical and thus very quick communicating visual formulation for his advertising message, he can introduce schematic pictorial elements in his/her picture. On the other hand, the danger of exchangeability increases which can, however, be warded off by the pictorial execution (= fourth module).

The knowledge on schemata and schema attributes delivered by the search system is obviously very fragmentary. However, if the user looks for schematic associations not available in the system, the stored schemata may serve him as paradigmatic examples to inspire his/her search for corresponding schematic visuals in other search fields.

The knowledge of intercultural differences of schemata and especially of visual schema attributes including the considerable cultural deviations in color schemata (see Jacobs, Keown et al., 1991) are important prerequisites for an effective message transfer in the different countries.

In short: To develop and evaluate an advertisement which can be launched in the different European countries, a very comprehensive data basis is needed offering information on the different cognitive and emotional structures and responses in these countries. Knowledge based computer programs are especially suitable to impart this information by stored cultural association norms and associative networks for key concepts of advertising. To complement the German association norms in the CAAS (Computer Aided Advertising System) of the Institute for Consumer and Behavioral Research of the University of the Saarland the collection of corresponding French and English associations is under way.

REFERENCES

Esch, Franz-Rudolf (1990), Expertensystem zur Beurteilung von Anzeigenwerbung (Expert System for the Evaluation of Printed Ads), Heidelberg: Physica.

Esch, Franz Rudolf and Werner Kroeber-Riel (1992), "Expertensysteme fnr die Werbung (Expert Systems for Advertising)", in Handbuch des Electronic Marketing, Arnold Hermanns and Volker Flegel (eds.), Mnnchen: C.H. Beck, S. 488-509.

Ger, Guliz and Russell W. Belk (1990), "Measuring and Comparing Materialism Cross-Culturally", in Advances in Consumer Research, XVII, Marvin E. Goldberg, Gerald Gorn and Richard W. Pollay (eds.), Provo, UT: ACR, 186-192.

Granowski, Gabi (1989), Unterschiede im emotionalen Gehalt der Werbung fnr Konsumgnter in den LSndern Italien, Frankreich, England und Deutschland (Differences in the Emotional Content of Consumer Ads in Italy, France, Great Britain and Germany), study submitted for a diploma at the Institute for Consumer and Behavioral Research of the University of the Saarlandes, Saarbrncken (Germany).

Grunert, Klaus G. (1990), Kognitive Strukturen in der Konsumforschung (Cognitive Structures in Consumer Research), Heidelberg: Physica.

Jacobs, Laurence, Charles Keown, Reginald Worthley and Kyung-Il Ghymn (1991). "Cross-cultural Colour Comparisons: Global Marketers Beware", in International Marketing Review, 8, Nr. 3, 21-30.

Kemper, Theodore D. (1991), "An Introduction to the Sociology of Emotions", in International Review of Studies on Emotion, Vol. 1, Ken T. Strongman (ed.), Baffins Lane, Chichester: J. Wiley & Sons, S. 301-350.

Keppel, Geoffrey and Bonnie Z. Strand (1970), "Free-Association Responses to the Primary Purposes and Other Responses Selected from the Palermo-Jenkins Norms", in Norms of Word Associations, Leo Postman and Geoffrey Keppel (eds.), New York und London: Academic Press, 177-240.

Kroeber-Riel, Werner and Franz-Rudolf Esch (1992), "Expertensysteme im Marketing (Expert Systems in Marketing)", in Handbuch des Electronic Marketing, Arnold Hermanns and Volker Flegel (eds.), Mnnchen: C.H. Beck, S. 250-269.

Lorson, Tania (1992), Entwicklung eines Expertensystems zur Beurteilung von Fernsehwerbung (Development of an Expert System for the Evaluation of Commercials), Heidelberg: Physica.

Maas, Jutta (1993), Visuelle Schemata in der WerbungBGrundlagen und Anwendung in einem computergestntzten Suchsystem zur Bildideenfindung (Visual Schemata in Advertising), Dissertation, University of the Saarland (Germany), Dept. of Business Administration (in preparation).

Petri, Carlo (1992), Entstehung und Entwicklung kreativer Werbeideen (Creation and Development of Creative Advertising Ideas), Heidelberg: Physica.

Postman, Leo and Geoffrey Keppel (eds.) (1970), Norms of Word Associations, New York und London: Academic Press.

Russell, Wallace A. (1970), "The Complete German Language Norms for Responses to 100 Words from the Kent-Rosanoff Association Test", in Norms of Word Associations, Leo Postman and Geoffrey Keppel (eds.), New York und London: Academic Press, 53-94.

Rosenzweig, Mark R. (1970), "International Ken-Rosanoff Word Association Norms, Emphasizing Those of French Male and Female Students and French Workmen" in Norms of Word Associations, Leo Postman and Geoffrey Keppel (eds.), New York und London: Academic Press, 95-176.

Stillings, Neil A., Mark H. Feinstein, Jay L. Garfield, Edwina L. Rissland, David A. Rosenbaum, Steven E. Weisler and Lynne Baker-Ward (1991), Cognitive ScienceBAn Introduction, Cambridge: MIT Press.

Szalay, Lorand B. and James Deese (1978), Subjective Meaning and Culture, An Assessment Through Word Association, Hillsdale, N.J.: Lawrence Erlbaum.

TTlTperformance (1988), Etude EuropTenne sur la ComprThension de l'Anglais, Lintas France (ed.), Paris: Lintas.

Watson, David L. (1989), "Enhancing Creative Productivity with the Fisher Association Lists", in Journal of Creative Behavior, 23, 1, 51-58.

Woesler de Panafieu, C. (1988), "Vision '92: Der Euro-Verbraucher, Soziokulturelle Zielgruppen als Basis globalen Marketings (The Euro-Consumer, Socio-cultural Target Groups as Basis of Global Marketing)", in Europa 1992, Nnrnberger Akademie fnr Absatzwirtschaft (ed.), Nnrnberg: NAA.

----------------------------------------