DOC.COM: a framework for effective negotiation support in ... - MV WEB

Computer Networks 37 (2001) 153±170

www.elsevier.com/locate/comnet

DOC.COM: a framework for e€ective negotiation support in electronic marketplaces Mareike Schoop *, Christoph Quix Informatik V (Information Systems), RWTH Aachen, 52056 Aachen, Germany

Abstract Today, research in electronic negotiations focuses on negotiation protocols that support the negotiation of only a few properties. These protocols are mainly designed for automated negotiations between software agents. However, for peer-to-peer negotiations in business-to-business electronic marketplaces, such protocols are less appropriate. For example, negotiations about multiple attributes, complex attribute combinations, or frame contracts require more sophisticated support of the involved interactions and communication acts. In this paper we present a novel approach to the e€ective support of electronic negotiations among human negotiators. Our approach is based on the following observation. The outcome of a successful negotiation process is a business contract. The contract evolves during the negotiation, coordinated through the exchange of structured messages such as o€ers, requests, quotations, countero€ers, and acceptances. Therefore, each negotiation between business partners involves the exchanges of documents and messages. Furthermore, documents and messages are interrelated in that a message leads to a new contract version which itself is the medium for a new negotiation step and thus initiates a new message. In contrast to the current practice of separate management of messages and documents, we create a powerful framework for e€ective negotiation support in electronic marketplaces by combining communication and document management. Our approach enables monitoring of contractual obligations and traceability of both documents and messages and their interrelations. Ó 2001 Elsevier Science B.V. All rights reserved. Keywords: Negotiation support; Electronic marketplaces; Communication management; Document management; Business-to-business electronic commerce

1. Introduction Electronic commerce (EC) has turned from a vision into reality. The area of business-to-business electronic commerce (B2B EC) has seen a growing interest both from researchers and practitioners. In the B2B arena, marketplaces often

*

Corresponding author. E-mail addresses: [email protected], http://www-i5.informatik.rwth-aachen.de/~schoop (M. Schoop), [email protected] (C. Quix).

provide a forum for customers and suppliers to conduct electronic trade. First of all, a customer looks for potential suppliers with respect to speci®c products, or a supplier wants to locate potential customers for the company's products. The marketplace thus provides the mechanism to search for products and/or companies. After locating potential (new) partners, the second step is to come to an agreement that is acceptable to all partners. In this phase, negotiation about contract details takes place. If a business deal is struck, the outcome is a contract which will then have to be processed by the partners, e.g., concerning

1389-1286/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved. PII: S 1 3 8 9 - 1 2 8 6 ( 0 1 ) 0 0 2 1 3 - 4

154

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

logistics, payment, etc. Therefore, we can extract the following general three-phase-model of a B2B EC process [21,35]: search, negotiate, and ful®l. The current practices can best be discussed using an example of an existing B2B marketplace of the chemical industry called chemUnity (www.chemUnity.com). A buyer's request containing information about the product, the type, concentration, delivery address and time is transferred via the marketplace to all potential suppliers. Suppliers have a ®xed amount of time (usually 25 h) to react. Those who choose to send an o€er will be taken into account. The best o€er is determined based on the buyer's selection criteria. If the best o€er is within the price range indicated by the buyer, then the transaction is completed and the following obligations exist: The seller must supply the product(s) indicated in the original request whereas the buyer must provide the payment according to the o€er received. In the present paper we will concentrate on the second phase of the commerce process, i.e., the phase of negotiation. Using again the example of chemUnity, a common approach to negotiation support is as follows. The buyer indicates a preferred price together with the highest price (s)he is willing to pay. This model is useful when automated agents take over the ``negotiation''. If negotiations are about one characteristic of the product, e.g., the price, then this approach can indeed be the most useful one. Likewise, negotiation protocols often focus on auctions [17,18] which can be implemented using the above model, be it about one or multiple attributes. However, complex negotiations are not possible with this model. For example, a negotiation could combine various characteristics such as price, delivery times, quality, supplier's location, means of transport, credit history, etc. Furthermore, negotiations often concern not just single bids but complex interactions as in traditional commerce. Finally, if negotiations concern frame contracts, complex interactions take place that cannot be covered by a simple model. In this case, human negotiators cannot and indeed should not be replaced by software agents and the aim should be to provide support for electronic negotiations rather than to automate them completely.

In this paper we will present an approach to negotiation support that is communication-based and interaction-based. Electronic negotiations involve the exchange of structured electronic messages and documents. The messages form the medium of communication among the business partners. In this paper, a novel framework of structured message exchange for communication management will be presented based on theories of communication and formal logics. Systems based on the framework will ensure ecient unambiguous interactions in which the obligations are clear for all business partners involved. In addition to messages, negotiations involve the exchange of documents. These documents are versions of the contract between the negotiating parties. During the negotiation process, information in the contract is changed which leads to a new version of the contract. Similar to the novel forms of communication support, a framework for document management is presented in this paper that ful®ls the requirements of electronic negotiations in B2B EC. For example, we argue for a semi-structured content since the content of documents exchanged during negotiations is not a ®xed set of features. Furthermore, it must be possible to discard a current version of the contract and reuse an old version. Thus, backtracking is an important requirement. Traditional document management systems support the evolution of documents by keeping track of di€erent versions. However, such systems do not provide facilities to track the messages which are exchanged during the evolution of a document. On the other hand, communication management systems manage the structure of messages that are exchanged but do not consider the documents that might be initiated by the messages. It is argued in this paper that there can be no separate document and communication management for e€ectively supporting electronic negotiations. Therefore, we propose to link the messages to the documents. A powerful framework for e€ective negotiation support in EC will be presented that combines communication support and document management. The structure of the paper is as follows. Before presenting the novel approach to communication

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

management for electronic negotiations, the relevant elements of two communication theories will be introduced in Section 2. The discussion of requirements for e€ective document management in that context follows in Section 3. The main contribution of the present paper is the combination of document management and communication support. Section 4 starts with the presentation of a conceptual model for electronic negotiations. We then argue for a representation of message and contract contents as extensible semi-structured documents. A formalisation of the framework DOC.COM will be introduced enabling reasoning about obligations and duties. The contents of obligations is linked to the contract and messages exchanged during the negotiation which allows queries about documents and messages. Thus, contract versions and messages can be accessed in many di€erent ways, e.g., through the message type, di€erent features of the contents (such as price, goods), sender, open obligations, etc. Section 5 presents further applications of our work. Finally, a discussion of the merits of our approach will conclude the paper (Section 6). 2. Communication management In this section, our vision of communication management for EC will be presented. Firstly, we will brie¯y introduce two theories of communication as the theoretical foundations. Using elements of these two theories, the structured message exchange as a means of e€ective negotiation support will then be described. 2.1. Theoretical foundations It has been reported that the main obstacle to smooth and e€ective cooperative interactions are fundamental communication problems and that communication is used for coordinating multidisciplinary activities [26,27,32]. Therefore, our idea was to look at communication theories as the theoretical foundations. The relevant theories for the present work are Searle's Theory of Speech Acts [33,34] and Habermas' Theory of Communicative Action [16]

155

which have both been in¯uential in the ®eld of information systems. 2.1.1. The theory of speech acts The Theory of Speech Acts was published by John Searle in 1969 [33,34]. Searle argues that the minimal unit of an utterance is not a word or a sentence but a speech act. Each speech act (such as ``I promise to deliver the goods before 4/8/2001'') consists of two elements: the propositional content describes what the utterance is about (deliver the goods before 4/8/2001) whereas the illocutionary force describes the way it was uttered (in this case as a promise). Each speech act has got a ``point'' which characterises that particular type of speech act. For example, an assertion is about informing other people, a request is about getting the recipient to perform an action, etc. This purpose of the act is called the illocutionary point. Searle classi®es utterances according to the illocutionary point into ®ve classes: assertives represent facts of the real world, e.g., reports; commissives represent the speaker's intention to perform an action, e.g., promises; directives represent the speaker's attempt to get the hearer to perform the action indicated in the propositional content, e.g., requests; expressives describe the speaker's feelings or psychological attitudes, e.g., apologies; declaratives change the world by their utterance, e.g., sentencing a prisoner. Each speech act implies a commitment for speaker or hearer. The most important forms of commitment occur in commissive and directive speech acts as they deal directly with the coordination of actions. A commissive speech act commits the speaker to perform the action (s)he indicated in the propositional content of the utterance. A directive speech act issues a commitment for the hearer to perform the action. Uttering the example speech act above would thus mean committing oneself to deliver the goods before 4/8/ 2001. 2.1.2. The theory of communicative action J urgen Habermas published the Theory of Communicative Action in 1981 [16]. He argues

156

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

that a speaker making an utterance makes four implicit validity claims: · the utterance is comprehensible so that the hearer can understand the speaker; · the utterance is true, i.e., it represents a fact or a common experience, so that the hearer can share the speaker's knowledge; · the utterance is truthful, i.e., the speaker is sincere in expressing his or her intentions, so that the hearer can trust the speaker; · the utterance is appropriate in relation to a given normative context, values, or standards, so that the hearer can agree with the speaker in these values. These four validity claims can be seen as four potential causes for communication breakdowns: If the utterance is incomprehensible then the speaker must rephrase or translate it. If the hearer challenges the truth of the speaker's utterance, then the speaker must be able to justify the utterance, e.g., by providing reasons, explaining it, supplying more information. If the truthfulness of an utterance is problematic then the speaker's intentions are questioned. Communication can only continue if the speaker succeeds in restoring the trust, e.g., through acting consistently, assuring the hearer of the speaker's sincerity, etc. If the appropriateness of an utterance is challenged then the hearer questions the speaker's right to perform the speech act, e.g., if a speaker's role does not entitle him or her to do so, if a speaker violates recognised values or acts contradictory to norms. These problems are usually solved by pointing to other (unproblematic) standards and norms, referring to common experiences, citing relevant literature or authorities. 2.1.3. The language-action perspective The broader context of the present work lies in the so-called language-action perspective (LAP). LAP is based on Searle's and Habermas' theories and focuses on communication aspects in information systems which is also important for the present work. This section presents the languageaction paradigm. The LAP ®rst introduced in the ®eld of information systems in 1980 [11]. It was stated that human beings are fundamentally linguistic beings

and act through language. It was argued that language is not only used for exchanging information as in reports, statements, etc., but also to perform actions, e.g., promises, orders, etc. The conventional perspective on information systems stresses the contents of messages rather than the way they are exchanged [20]. For example, data ¯ow diagrams are used as primary design tools. Thus, the focus is on the form and structure of messages. In contrast, the LAP emphasises what people do while communicating, how language is used to create a common reality for all communication partners, and how their activities are coordinated through language. Here, the focus is on the pragmatic aspect of language, i.e., how language is used in particular contexts to achieve practical goals such as agreements or mutual understandings. This new approach argues that as social action is mediated through communication, the main role of an information system should be to support organisational communication. LAP has since developed into a new paradigm for the design of computer systems. There are a number of basic assumptions underlying LAP [19,39], e.g.: · The basic unit of communication is a speech act. · Natural language sentences correspond to the performance of speech acts. · The meaning of sentences can be revealed by specifying the speech acts that have been performed. · Speech acts obey socially determined rules. · Cooperative work is coordinated by the performance of language actions which are speech acts. The early work on LAP is based on Searle's Theory of Speech Acts. As a result of criticism of the shortcomings of Searle's theory (e.g., [6,37]), Habermas' Theory of Communicative Action is nowadays combined with Searle's theory as the philosophical foundation of LAP. Although both Searle and Habermas talk about ``speaker'' and ``hearer'', their theories can be applied to written communication as well [32]. In the following section it will be discussed how LAP and the two communication theories in particular in¯uenced our approach to e€ective communication management for electronic negotiations.

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

2.2. Structured message exchange We argue that communication management plays an important role for an e€ective support of electronic negotiations. During negotiations, the partners communicate with each other. In traditional forms of negotiations, the communication can take place directly (i.e., face-to-face), via telephone, using letters or faxes, sending e-mails, etc. If negotiation is to take place electronically, the forms of business exchange will be primarily e-mail messages. On the one hand, this enables distributed work as the negotiation partners do not need to communicate with each other at the same time (as, for example, in telephone or face-to-face conversations). It is possible to reach new business partners all over the world. Furthermore, other people that are not the main negotiators but that are involved in some way (such as lawyers or ®nancial advisors) can be kept informed easily (e.g., by using cc. in e-mails). Finally, e-mail exchanges can be logged so that there is the possibility to trace back business processes. On the other hand, there are a number of potentially severe problems concerning e-mail messages. E-mail is one form of written communication. It has been reported that communication problems which can easily be solved in face-to-face interactions, can become serious in the case of written communication [25]. If such problems are not solved, then communication can break down completely which in turn can lead to breakdowns in cooperation [28,32]. Therefore, it is of prime importance to anticipate potential communication problems to ensure unambiguous, structured, and ecient message exchange as the basis for the support of electronic negotiations. Next we will present features an ideal communication management system for electronic negotiations would contain. Elements from the two theories of communication introduced in Section 2.1 will be used and extended. The following scenario will illustrate our approach throughout this paper: Company A wants to buy shoes with certain characteristics (leather sole, black colour) from company B. Let us imagine that a buyer of company A contacts a sales person

157

of company B by electronic means. They will start a negotiation about the contract details. 2.2.1. Message type To ensure unambiguity, the message type needs to be speci®ed. A negotiation process starts either with a request or an o€er made by the customer and supplier, respectively. A supplier needs to know whether the message concerning certain goods (s)he receives is meant as an order of these goods or as a mere inquiry without obligations. Furthermore, reading only the content of a message might leave it unclear whether it is meant as a request to buy something or as an o€er by a certain supplier wanting to sell the goods. Referring to the scenario, a statement such as ``100 shoes for £25'' could be interpreted as a request for quotation or as an o€er. Therefore, the type of each message (which indicates the illocutionary force) needs to be made explicit. This will also make the state of the negotiation process clear for the communication partners, i.e., whether they are in the inquiry phase or in the commitment phase. It has been shown that Searle's classi®cation into ®ve types of illocutionary force (i.e., assertive, commissive, directive, expressive, declarative) is useful for specifying the types of messages in the present context [27]. 2.2.2. Message content Speech act theory and its formalisation, i.e., illocutionary logic [34], treat the propositional content as primitive. We argue that a formalisation of the message content needs to take place to ensure unambiguity of the message contents. In the example, the content of the message might be misunderstood. The exact speci®cation of the product or the quantity (100 shoes or 100 pairs of shoes?) might be problematic. On the one hand, the content of messages needs to be speci®ed to enable queries such as ``Which goods do we need to deliver to company A?'' or ``What did company B o€er?'' that concern the propositional content. On the other hand, systems that only o€er prede®ned message contents appear too rigid and in¯exible. Therefore, semi-structured message contents should be aimed for.

158

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

2.2.3. Validity claims Validity claims as introduced by Habermas play an important role in negotiations. Validity claims are critical to the success of speech acts. Only if the recipient of a speech act says implicitly ``yes'' to all the claims raised, can the speech act be successful. The consideration of validity claims allows fruitful discursive communication. We argue that the recipient of a message needs to have the opportunity to challenge the claims made by the sender. To allow only sensible challenges, a system for negotiation support should present the recipient with the relevant claims for the particular type of speech act. We will brie¯y summarise the relations between the validity claims and the components of the speech act (i.e., propositional content and illocutionary force) for all types of act, see Table 1. The interested reader is referred to [26,28] where an extensive discussion is presented. An utterance can be incomprehensible because the propositional content is not understood by the recipient. This is a common communication problem. However, it is also possible to challenge the comprehensibility of the illocutionary force used in a speech act. Illocutionary forces are not always speci®ed in an utterance which can lead to misunderstandings about which force the sender meant to use when uttering a sentence, e.g., what was meant as a request is understood as a mere statement. Questioning the truth of a statement means questioning whether the statement really represents a fact, a common experience etc. Thus, the validity claim of ``truth'' is related to the propositional content only. The challenge of truthfulness is only related to the illocutionary force since the recipient doubts Table 1 Relations between validity claims and speech act components Validity claim

Comprehensibility Truth Truthfulness Appropriateness

Speech act component Illocutionary force

Propositional content

+

++ +

+ ++

+

whether the sender is really committed to the illocutionary force used. Here, the ostensible nature of the speech act is questioned, e.g., what seems to be an assertive act is really an attempt to deceive. Both the illocutionary force and the propositional content can be inappropriate. However, a challenge of appropriateness concerns more often the illocutionary force than the propositional content. The recipient challenges the appropriateness of the force used by the sender by questioning whether the sender is entitled to use the force in the particular context. The sender could have violated existing power relations, recognised norms, or professional standards. A propositional content can also be inappropriate. Here, the recipient questions whether the sentence ®ts a given normative context, i.e., whether it is legitimate to say so. The relations between validity claims and the ®ve types of speech act have also been analysed, see [26,28]. To summarise, comprehensibility problems can occur in all utterances, the claim of truth is only related to assertive speech acts, truthfulness can be challenged for all types of utterance, and appropriateness is related to assertive, directive, expressive, and declarative speech acts. 2.2.4. Order of messages The order of a negotiation process should be speci®ed to guide the actual business interactions. For example, a negotiation can start with a request by the buyer or an o€er made by the seller. In the ®rst case, the seller can accept the request, can reject it, or can reply with a countero€er, see Fig. 1. Furthermore, there can be a discussion request based on the problematic validity claim(s). This indicates stepping out of the normal negotiation order and shows a kind of ``time-out'' in which problems are solved. In general, negotiation steps cannot be done in a random order. For example, a countero€er of company A cannot be followed by another countero€er or a request of company A, because only company B is allowed to respond with a countero€er or to accept or reject the o€er. Because of this, in conventional negotiations it is sometimes unclear which company has to react. Therefore, the possible answers to certain message types need to be speci®ed.

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

159

2.3. Summary

Fig. 1. Order of negotiation steps.

2.2.5. Obligations As mentioned in Section 2.1.1, directive and commissive speech acts issue commitments for the recipient and the sender, respectively. For example, an inquiry (which is a directive speech act) commits the recipient to answer. A request made by a buyer and accepted by a seller implies that the seller is obliged to perform the action indicated in the message. Once a speech act is accepted (cf. Section 2.2.3), an obligation arises. It needs to be speci®ed what kind of obligations the companies have already accepted during a negotiation. To ensure that all business partners know their duties, the obligations resulting from the exchanges need to be made explicit. 2.2.6. Traces An important requirement of electronic negotiations is that the exchanges should be logged to allow backtracking and traceability. In a messageoriented view, the di€erent messages that were exchanged are linked. Any message (apart from the one starting the negotiation) is a reply to one message; any message (apart from the one terminating the negotiation) can have many answers. The temporal sequence of messages can be displayed in a tree-like structure. Being able to trace messages of a particular negotiation process provides many advantages, e.g., enabling the evaluation of di€erent negotiation strategies, providing a reminder of what has already taken place in a negotiation process, telling other parties involved about important information, providing the basis for monitoring mechanisms in case of later con¯icts.

In this section we have presented our speech act-based approach to e€ective communication management. Work in the context of information systems in¯uenced by speech act elements has been done for a number of years [7,12,13,31] including work on negotiation exchanges, e.g., [10]. The knowledge query and manipulation language (KQML) provides a protocol for exchanging information and knowledge for automated software agents. It is, on the one hand, very formal by providing technological protocols for agent interactions while, on the other hand, it is not rigorous enough to supporting human negotiations. For example, there is no (formal) representation of the message content. Therefore, such a language is not suitable for the present purpose. So far, the basis for the e€ective support of human communication in electronic negotiations has been laid. Apart from structured electronic messages, the other important component of electronic negotiations are documents that are exchanged during the negotiation process. The following section will present our approach to document management for the context of electronic negotiations between human agents. 3. Document management for electronic commerce The previous section introduced our work on communication management. However, the negotiation process between business partners does not only involve the exchange of messages. Documents such as contracts and general business conditions are also important. Thus, the management of documents during a negotiation is also necessary. In this section, we will present our vision of e€ective document management for EC by specifying features of ideal systems. 3.1. Contract management By de®ning a contract, the business partners work cooperatively on a document. The contract evolves over time until a ®nal agreement has been reached or the negotiation is terminated. There-

160

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

fore, documents are both the medium and the outcome of negotiations. During the negotiation process, a new version of the contract may be a dead end. In this case, the partners will go back to an earlier version of the document. Thus, the system must be able to manage di€erent versions of a document and reuse of an old version must be possible. The version management can also be used to trace the evolution of the document. For example, the versions of a document can be displayed in a tree-like structure with the initial version as the root and the revised documents as successors. Business documents, especially contracts, do not have a ®xed structure. The structure evolves during the negotiations, e.g., items are added and removed. Because of this, the management in a traditional relational database system would be very inecient. On the one hand, the document could be stored as a binary object without knowing the details about the contents. In this case, one can only search for keywords in these documents. On the other hand, each item of the document could be stored as an attribute of a relation. However, as the structure of the document evolves, the database schema needs to be changed as well. Thus, the database schema might be changed very often so that it becomes dicult to have an ecient database design. 3.2. Semi-structured data A more ecient way to handle documents is to view them as semi-structured documents. A document-oriented database system such as Lotus Notes [23] manages semi-structured documents: some parts of the document can be identi®ed as ®elds (similar to attributes of a table in a relational database), but the document may also contain unstructured elements such as text and images. The structure of documents is not described in a schema. Therefore, each document can have a di€erent structure and may adjust its structure as needed. This has also been pointed out by [1]. It is not sucient to store a document in a binary ®eld within a database. The internal structure and semantics of a document also need to be considered

to enable more ecient document retrieval and processing techniques. Recently, XML (eXtensible Markup Language) has become a popular standard for representing (semi-)structured data [4,38]. Basically, XML is a syntax speci®cation for information exchange between di€erent tools. The advantage of XML is its ¯exibility and extensibility, while at the same time being simple and easy to understand. Therefore, XML has been used by a huge number of commercial products in the area of electronic commerce. XML documents have a tree-like structure. A document has one root element which can have multiple elements as successors. These elements can again have multiple elements as successors and so on. Each element is marked by a tag. For example, ``hprodcodei MM-48-1112 h=prodcodei'' is the tag for a product code with the contents ``MM-48-1112''. The structure of an XML document can be described in a Document Type De®nition (DTD). A DTD speci®es the tags which may appear in the document (e.g., hprodi and hprodcodei). Furthermore, it describes the order (or structure) in which the elements of an XML document may appear. The set of tags is not restricted in contrast to HTML. New tags can be de®ned as they are needed. A DTD is not necessary for an XML document. However, for data exchange between di€erent application programs, the applications must agree on the same DTD. There is some standardisation e€ort in di€erent industry sectors going on to de®ne DTDs for their application domain. 3.3. Document management systems Some of the functionality discussed before is o€ered by document management systems [15]. Document management systems are used to support the work¯ow management within an organisation. In this environment, documents are containers of information which can be read or ®lled in by a group of people in a coordinated way. Such a functionality is particularly well supported by systems such as Lotus Notes or Basic Support for Cooperative Work (BSCW, [2]) that combine

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

the features of document management systems and e-mail systems. However, the focus of such systems is obviously on document management and thus their support for communication management is rudimentary and not well integrated with the document management system. For example, Lotus Notes as a groupware system o€ers also an ``integrated'' e-mail system. Links to documents can be sent to other group members by e-mail. As the document resides at a central place, the user is always working on the latest version of the document. Furthermore, Notes supports work¯ows by noti®cation and monitoring mechanisms. For example, an e-mail with a document link can be sent to the next responsible employee when a certain event occurs. If the employee does not react within a certain time, the system can send an e-mail to the supervisor to take care of the problem. Other document management systems o€er similar functionalities for work¯ow support or workgroup collaboration. Another focus of document management systems is cross-media publishing, content management, and catalogue management. Documentum (http://www.documentum.com) is a system aiming at supporting these tasks in a combined way. 3.4. Summary So far, the two main components of electronic negotiations, i.e., documents and structured messages, have been introduced. In this section, our approach to document management was discussed while Section 2 presented the ecient management of communication. Both types of management are interrelated in electronic negotiations since negotiation processes involve the exchange of structured messages that in turn lead to new contract versions. Therefore, a combination of communication and document management is essential. The way for both parts has been paved in the present section and the previous one. In the following section, their combination is presented which provides the basis for system support for human negotiators in their complex tasks.

161

4. DOC.COM: Combining document and communication management Although some of the functionality required for managing messages and documents in B2B negotiations is available in individual systems, the combination of messages and documents is essential. Traditional document management systems support the evolution of documents by keeping track of di€erent versions [15]. However, such systems do not provide facilities to track the messages which are exchanged during the evolution of a document. On the other hand, communication management systems are not concerned with documents that might result from message exchanges. We argue that there can be no separate communciation management and document management if the goal is e€ective support of electronic negotiations. Both messages and documents are an essential part of the negotiation process and both are interlinked. An electronic negotiation starts with a message sent from one company to the other. The type is either a request or an o€er. In either case, a new document is derived from the message. The document is the ®rst version of a contract between the business partners. In the course of the negotiation, new documents (i.e., new contract versions) are created through the message exchange between the negotiators. In our approach, the messages are linked to the documents, thereby combining document management and communication management. We do not consider a message as a subtype of a document because the relationship among messages is of a di€erent type than the relationship among documents (replies vs. versions). Furthermore, the content of messages should be related to the content of documents to indicate why a certain information is included in a contract version. The combination of document and communication management (DOC.COM) is the main emphasis in this paper and will be presented in this section. To prepare for this work, a conceptual model to capture the relationship between documents and messages will ®rst be introduced. It will then be shown how the content of negotiation processes

162

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

can be represented in a more formal way to enable reasoning about their consequences. 4.1. A conceptual model for electronic negotiations Our conceptual model of document management in electronic negotiations is shown in Fig. 2 in entity-relationship notation. The context of business interactions is a negotiation which involves two or more parties and has a certain subject. Negotiations can be active or terminated which is represented by the attribute ``status''. A negotiation contains documents as well as messages. A message is sent by one of the business partners to the other partner. It is sent at a certain time and has a semi-structured content which represents the propositional content as described before. We will elaborate in more detail on the representation of the content below. Furthermore, the message type needs to be speci®ed (e.g., whether it is a request, a countero€er, an assertion, etc.) which represents the illocutionary force. As messages are sent among the business partners involved in the negotiation, they are linked in a content-based sequence. When sending a message, the business partner answers a message sent by the other partner (cf. Section 2.2.4). One message can of course have many answer messages. Imagine the situation where a request for products with a certain price and a due date is sent. The seller might send a message agreeing on the price whilst having to talk to the production manager about the due date. Later on, a new message is sent answering

part of the original request which in this case would constitute a reply to the proposed due date. A document has a status which speci®es whether it is the ®nal contract or an evolving version of the contract. The content of the document is the subject of the negotiation steps. As we will discuss below, the content of a document is also represented in a semi-structured form. Documents can be ordered in a tree-like structure. Thus, there is a many-to-many relationship called ``is successor of'' between documents. It is possible to turn back to a certain earlier document (say document version 5) during the negotiation process and to create a new version of the contract (i.e., a new successor of the document version 5, say document version 12) from there. Thus, it is necessary to keep track of the time of the relationship. Document version 5 had only document version 6 as a successor before, but at a later point in time, document version 12 also became the successor. Thereby it is possible to keep the documents in a temporal sequence (cf. the discussion of version control in Section 3.1). Messages and documents are linked because messages initiate new documents, i.e., new versions of the contract. For example, a request made by company A to company B of 100 pairs of shoes of code MM-48-1112 for £20 each is a message that creates a document (i.e., a contract version) with the content of ``B sends 100 pairs of shoes of code MM-48-1112 for £20 each to company A''. This document then leads to further discussions through the medium of structured message exchange.

Fig. 2. The conceptual model of electronic negotiations.

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

In general, the contents of the document can be derived partly from the contents of the messages. If a partner proposes another price or a new delivery date in a message, this item can be inserted or updated in the contract document. To do so, we must be able to identify certain elements of a message or a document. The conceptual model provides only the relationship between messages and documents, but relationships between speci®c items of a message or document cannot be captured by this model. Therefore, a more structured representation of the contents is necessary, because the model alone does not provide information about why a data item in a contract has been inserted. The idea of DOC.COM is to represent the contents of messages and documents as extensible semi-structured documents. The idea is based on the observation that a contract or a message consists of several items which are important to identify. For example, a statement such as ``B sends 100 pairs of shoes of code MM-48-1112 for £20 each to company A'' contains information about the selling and buying companies, the quantity, the product, the product price, and the product code. These information items should be marked in a message or document as items with a special meaning. This can be done by using XML. Each element of interest can be tagged with a special tag as XML is extensible. Continuing the example, a contract between company A and company B can be drawn up in the following way: hcontracti hdeliveryi hsupplieri B h=supplieri delivers to hrecipienti A h=recipienti hquantityi 100 h=quantityi huniti pairs h=uniti of hproducti shoes h=producti of code hprodcodei MM-48-1112 h=prodcodei until hdatei 10/03/2001 h=datei h=deliveryi hpaymenti hpayeri A h=payeri has to pay to hrecipienti B h=recipienti hamounti 2300 h=amounti hcurrencyi GBP h=currencyi within a

163

htermi month h=termi after delivery h=paymenti h=contracti The example speci®es a contract between companies A and B. The ®rst part describes the obligation that company B has to deliver shoes within a certain time to company B. If B has delivered the goods, A has to pay £2300 to B within a month. The individual obligations of a company can now be identi®ed as a part of the contract. For example, the obligation of company B to deliver shoes to company A can be represented as the path expression contract.delivery. Individual items of the document can also be identi®ed by path expressions, e.g., the delivery date is identi®ed by contract.delivery.date. Items can now also be linked to the messages in which they were proposed, thus giving information about who proposed this item in which context. XML representations of documents have been used in many approaches. An example in the context of electronic negotiations is [36]. In an implementation of DOC.COM, we follow a di€erent, but semantically equivalent, approach. The contents of the message is entered as free text. Parts of the text can be marked and linked to an ontology of concepts. The ontology of concepts is organised in a hierarchical structure and represents contract elements such as products and their attributes, delivery date, etc. Based on our argumentation that the structure of the content of messages and documents cannot and should not completely be pre-de®ned, it is possible for the negotiator to extend the ontology by marking text as the value of a new user-de®ned element. The relationships between the text elements and the ontology concepts are stored as attribute-value pairs. This enables ecient querying of messages with a certain content, e.g., all messages that refer to product MM-48-1112. The screenshot in Fig. 3 shows an example. The ®eld of free text contains an element ``280,DM'' which the author marked and de®ned as the price. Thus the element ``280,- DM'' is linked to the concept price, shown on the left hand side. As an element is marked in the message, it is inserted or updated in the corresponding docu-

164

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

Fig. 3. Creation of a message and marking of elements.

ment. Certain elements of the ontology need to be speci®ed in every contract, e.g., buyer and seller, product, price, delivery date, etc. As these items are now made explicit, we can check whether they have been de®ned or not. Thus, the system supports the negotiator in de®ning and checking a contract. Based on the formalisation presented in the following section, more complex constraints and consistency checks for negotiations can be implemented. 4.2. Formalisation of the framework So far, the framework of combined document and communication management has been introduced on a conceptual level. Here, we will brie¯y present a formalisation which provides the basis for reasoning mechanisms, e.g., about arising obligations. A language coo L (standing for cooperative language) has been developed in [26]. It is based on a combination of dynamic deontic logic [22] and illocutionary logic [34], originally devel-

oped in [8,9,39] but considerably extended and enhanced for the present purposes, namely communication partners coordinating their work by exchanging speech acts about certain actions or propositions in a particular temportal order. The language enables statements about speech acts concerning actions or propositions to be made by an author to a recipient; it enables the resulting e€ects such as beliefs or obligations to be speci®ed; and it allows deadlines for certain types of speech act to be made explicit. We will present some elements of the formalisation in the usual modal-logic notation. Ora …a† stands for: r is obliged to a to perform action a; ‰aŠ/ stands for: / holds after the performance of a; (a1 ; a2 ) stands for: sequential performance of a1 and a2 . The illocutionary forces relevant for the present context are classi®ed into the ®ve categories: assertives, commissives, directives, expressives, and declaratives as described in Section 2.2.1. This is a general classi®cation, i.e., not depending on particular authors and receipients. Regarding the

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

acts themselves, the reader should remember that there are acts of acceptance and refusal made by the recipient of assertive, directive, expressives, and declarative speech acts (cf. Section 2.2.3). Thus, the force acceptDIR stands for the illocutionary force of accepting a directive speech act, i.e., accepting the validity claims that the author of this act has raised. With this in mind, an extract of the classi®cation of message types (representing illocutionary forces) in this case for the set of commissives forces COM for B2B EC is as follows: accept; evaluate; offer; promise; review; acceptDIR; refuseDIR 2 COM: Each class of utterances is related to certain obligations. However, only if the recipient explicitly accepts the author's utterance in the particular context can the obligation arise (cf. Section 2.2.3). This can be expressed by, for example ‰bar …a†; acceptDIRra …a†ŠOra …a† with b representing a directive, a an action, a, r human agents (representing author and recipient) which means that the issuance of a directive, followed by the recipient's acceptance of this act leads to an obligation for the recipient to perform the action. Since obligations only hold for a certain time (i.e., until they have been ful®lled or violated in case of missed deadlines), temporal operators are included into coo L. If an obligation exists, it is possible to specify deadlines. For example, an action might have to be carried out before a certain date or before a certain proposition becomes true. It is also possible to specify that the action needs to be done immediately or between two dates. The language coo L allows the obligations to be checked at any point in time. It is possible to query the existing obligations, the ones that have already been ful®lled, and the unful®lled obligations. If an obligation has not been ful®lled and the deadline is past, then the obligation no longer exists because it could never be ful®lled anymore. However, it is clear that there exists an unful®lled obligation. This could, for example, lead to penalties, reminders, automated actions, etc. coo L makes it possible to reason about ideal and actual behaviour and to initiate actions or procedures if vio-

165

lations occur, i.e., if the actual behaviour is not the desired one. For example, the violation of an obligation to perform an action a before a certain proposition / (which could, e.g., specify a date) becomes true could trigger another request by the same author to the same recipient to perform the action. This can be expressed by the following statement which will be explained intuitively below: …/ ^ :DONE…a† ^ P …Ora …a < /††† ! Oar …remindar …a††: The recipient was obliged to do a before / (represented by Ora …a < /†), the action has not been done as the previous action …:DONE…a†† and the obligation existed in the previous state (to ensure that a had not been done previously; represented by …P …Ora …a < /†††. This triggers the obligation for the author of the original act to perform a reminder. Rather than specifying the semantics of the language we will now give some examples to illustrate it. Ba …/† stands for: a believes in proposition /; Ora …/† stands for: r is obliged to a to bring about /; the other notations are used as before. Coming back to our scenario, imagine the following interaction. Mr X of company B and Ms Y of company A start a negotiation about shoes. Mr X states that B has reduced the prices within the last month (represented by /1 ). If Ms Y accepts this assertion (i.e., accepts the validity claims associated with it) she now believes in the statement. ‰stateXY …/1 †; acceptASSYX …/1 †ŠBY …/1 †: Ms Y makes a request to buy 100 pairs of shoes for £20 each (represented by a1 ). Mr X refuses. Therefore, no obligation arises. ‰requestYX …a1 †; refuseXY …a1 †Š:OXY …a1 † Mr X now makes a countero€er: Company B could o€er a price of £23 each but could deliver the shoes (represented by a2 ) before 10/03/2001 (represented by /2 ). Ms Y accepts this o€er and thus the obligation to deliver the shoes for the price speci®ed before the deadline (represented by OXY …a2 < /2 †) arises for Mr X.

166

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

‰offerXY …a2 †; acceptCOMYX …a2 †ŠOXY …a2 < /2 †: If the obligation is violated (i.e., the shoes have not been delivered in time) certain actions can be initiated. For example, a ®ne could have to be paid. This can also be expressed in coo L. Instead of representing the actions and propositions (ai and /i ) as natural language expressions, we can use the semi-structured representation introduced in Section 4.1 and refer to the contracts and messages. In our example, the obligation to perform a2 , i.e., company B has to deliver shoes to company A, can be represented by the path expression contract.delivery. The proposition /2 is expressed as contract.delivery.date. 5. Applications of DOC.COM The approach presented in this paper provides a framework for a communication-based and document-based negotiation support system for B2B EC. The conceptual model introduced in Section 4.1 provides the basis for this work. In this section di€erent contexts of application will be discussed to show the merits of our work. In traditional commerce, most of the documents and messages exchanged during a transaction are kept for evidence in case of con¯icts. Furthermore, o€ers and invoices from previous transactions contain valuable information, e.g., they might be useful for pre-calculations and cost estimates for new projects. In our approach, the messages and documents that are exchanged during a negotiation are logged. Therefore, the documentation that is so important in any (electronic) negotiation is done automatically. Documentation of the exchanges serves di€erent purposes, e.g., as a memory aid (negotiations can take place over a long period of time and in parallel; the negotiator can see at one glance what (s)he did last time, what the negotiation partner answered, etc.), as the basis for statistical applications (how many exchanges took place before an agreement was reached, was that particular strategy successful, etc.), and as a medium for communication with others (the records can be used by other members of the same company to evaluate their own strat-

egy; they can automatically be transferred to the relevant departments that are also involved such as the ®nancial department; the records are the medium of communication between the negotiators; the general records can be annotated by personal remarks of each negotiator that is only visible to that party) [26]. Since the proposed message and document exchanges are structured they provide the basis for e€ective tracing. Not only the outcome of an agreement is documented but also the history behind it which provides important additional information. The history of the contract consists of the message and the document history. The history of documents shows how the contract as the ®nal document evolved, which elements were agreed on from the beginning, which contract clauses were under negotiation, how the agreement on these clauses was reached, whether it was a straightforward negotiation or whether earlier document versions were re-used in the course of the interactions, etc. The history of messages shows the negotiation cycle of the involved o€ers, countero€ers, requests, etc. This is enabled by the formalisation of the messages type representing the illocutionary force. Furthermore, the history of messages can be used to ®nd out which partner included a certain item in the contract and the exchanges leading to it. In addition, the messages might give more information about the reason for a speci®c contract clause and clarify ambiguous statements. The trace of the negotiation can, therefore, be represented in a document-oriented view (i.e., a graph-like structure that shows the evolution of a document) or in a message-oriented view (i.e., a tree-like structure of message threads), see Figs. 4 and 5. The ®nal document of a successful negotiation, i.e., the contract, formalises the obligations of each partner. The formalisation of the framework provides a formal representation of such duties and thus enables to monitor the ful®llment of the contract. Having reached an agreement does not necessarily mean that all obligations will be ful®lled by the negotiation parties. Therefore, a trustworthy monitor could help to observe the interactions and to warn of potential breaches of the agreement. The monitoring role could be taken

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

Negotiation with Company B about 100 Shoes Document 0.1 Request for Quotation about 100 shoes to company B Date: May 2, 2001, View Related Messages Document 0.2 Offer about 100 shoes of company B Date: May 3, 2001, View Related Messages Document 0.3 Rejection Date: May 3, 2001, View Related Messages Document 0.2.1 Offer about 100 shoes of company B Date: May 4, 2001, View Related Messages Document 0.2.2 Counter-Offer Date: May 5, 2001, View Related Messages Document 0.2.3 Counter-Offer of company B Date: May 5, 2001, View Related Messages Document 1.0 Contract Date: May 8, 2001, View Related Messages

Fig. 4. Document-oriented view.

Negotiation with Company B about 100 Shoes Date: May 2, 2001 From: CompanyA To: CompanyB Type: Request Quotation Subject: 100 Shoes Content Document Date: May 3, 2001 From: CompanyB To: CompanyA Type: Offer Subject: 100 Shoes Content Document Date: May 3, 2001 From: CompanyA To: CompanyB Type: Reject Subject: 100 Shoes Content Document Date: May 4, 2001 From: CompanyB To: CompanyA Type: Offer Subject: 100 Shoes Content Document Date: May 5, 2001 From: CompanyA To: CompanyB Type: Counter-Offer Subject: 100 Shoes Content Document Date: May 5, 2001 From: CompanyB To: CompanyA Type: Counter-Offer Subject: 100 Shoes Content Document Date: May 8, 2001 From: CompanyA To: CompanyB Type: Accept Subject: 100 Shoes Content Contract Fig. 5. Message-oriented view.

over by a so-called trusted third party (TTP). There are many models of monitoring a negotiation, see [30]. The most obvious role would be to

167

monitor the ful®llment of a contract. Reasoning about obligations can be done based on the formal language coo L as discussed in Section 4.2. In case of con¯icts the TTP can trace back the exchanges that led to the agreement and can ®nd out where the problem lies and which party is in breach of the contract. Companies might prefer such an institution before going to court about contractual disagreements. Grimm and Ochsenschlager [14] present one possible formalisation of monitoring contractual obligations in the ful®llment phase. Thus, our approach provides the basis for such monitoring by making the obligations explicit. As our model provides the basis for a document and message data warehouse, another application of our approach is to ®nd out strategies for successful negotiations. Di€erent strategies can be compared and evaluated. Data mining techniques can be used to detect relationships and similarities of successful negotiations. Furthermore, the data of negotiations can be linked to other electronically stored information. For example, the information about the negotiation for purchasing a product can be linked to an entry of an enterprise resource planning (ERP) system that records the product when it is delivered. If a problem regarding the quality of the delivered product occurs, the messages of the negotiation can be traced back to ®nd out whether the supplier made any statement or promise about the product quality. Using the formal representation of documents, messages, and obligations and the link to the ontology of business terms, we are able to provide access to the content of a negotiation by a variety of di€erent means such as: · content, e.g., show all messages concerning bricks; show all documents dealing with shoes; · type, e.g., show all o€ers of company B; what are the requests I sent to Y? · people involved, e.g., show all messages sent by X; what are the documents changed by Z? · duties, e.g., what are the open obligations of our company? which obligations did my business partners not ful®l? · time, e.g., show all request sent to a company before 1/6/2001; which documents did I receive between 7/6/2001 and 11/11/2001?

168

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

6. Conclusion Negotiation is an essential component of EC processes. Traditional forms of negotiation have long been the subject of research concerning different strategies, policies, communication patterns, etc. Electronic negotiations provide a challenging research topic as, on the one hand, they o€er many new possibilities for the negotiators but, on the other hand, they remove the direct interpersonal relations that seem to be inherent in many negotiation processes. In general, negotiation is a complex communication process and involves the exchange of documents. In electronic negotiations, communication takes place in a written form through message exchange and the documents involved are versions of the ®nal contract. Both documents and messages are, therefore, essential elements of such interactions. We have presented an approach to support negotiations in B2B EC more eciently by combining communication and document management. The approach is based on communication theories which provide the framework for a structured message exchange. We have identi®ed several features of an ideal communication system. The medium and outcome of a negotiation is a contract. We pointed out that a version control and a semi-structured representation for contracts is necessary. The main contribution of this paper is the framework DOC.COM in which all aspects of communication and document management are integrated. Messages can be exchanged in a structured way. The document management is enhanced by the combination with the communication about the documents. By combining messages and documents we can provide e€ective support for all aspects of negotiation. It is important to point out that we do not expect users to learn Searle's and Habermas' theories before being able to use systems based on the framework. Rather, such systems would need a sophisticated user-centred design to acquire the necessary data without causing more time and e€ort for the users. Electronic negotiations have been the subject of many research approaches. In the present journal, four articles deal with negotiations and each article

emphasises di€erent issues. Bichler and Segev [3] present work on multi-attribute auctions; Choi et al. [5] focus on agent-based negotiation systems. In both approaches, the focus is on automated negotiations rather than on support for human negotiators. Grimm and Ochsenschlager [14] present a formalisation of contractual obligations in electronic commerce. In contrast to their approach, it is not our aim to prescribe the negotiation exchanges completely. The structure of the message exchange is prescribed in our work by providing the possible order of message types. The individual form of exchange and the content of the messages is speci®ed by the human negotiators since both are highly context-dependent. It is important to state again that our aim is not to automate negotiations but to support human negotiators in their complex negotiation tasks. An implementation of DOC.COM has been created for the area of construction projects and in particular for the negotiations between architects and di€erent trades such as roofers, bricklayers, window manufacturers, etc. The implementation and, more importantly, the underlying framework DOC.COM has been successfully evaluated in such negotiation scenarios. Therefore, DOC.COM is not merely a conceptual framework but a guide and foundation for the development and implementation of negotiation support systems. In this paper, we have focused on one phase of a B2B electronic commerce process, i.e., electronic negotiations. Our work is embedded in a holistic approach to supporting B2B interactions in the context of EC [24]. Our future work will focus on other application areas (such as electronic marketplaces for software components [29]) to determine the di€erences and the similarities and thus distinguish between the general features of our approach and the adaptions necessary for particular application contexts. Furthermore, we will combine the formal negotiation as introduced in this paper with an informal message exchange to re¯ect the fact that negotiations consist of informal interactions (where the two parties try to ®nd out whether they want to engage in serious negotiations) and formal exchanges (which re¯ect the serious negotiations between the business parties). Research focuses on

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170

deriving the obligations of each phase automatically. For example, an informal inquiry obliges the recipient to answer whereas a formal request obliges the recipient to accept, reject, or countero€er. Our future aim is to o€er a negotiation module in an electronic marketplace that consists of many models for di€erent contexts, e.g., the one we have presented in this paper, auctions, agent-based negotiations, etc. Therefore, we will extend the negotiation functionality of existing electronic marketplaces and o€er complete negotiation support in many facets that is suitable for di€erent contexts. Acknowledgements The research is supported by the European ESPRIT Project ``MEMO: Mediating and Monitoring Electronic Commerce'', No. 26895 (http:// www.abnamro.com/memo/). We would like to thank our project partners from CentER AR Tilburg for the useful discussions. The architecture practice Linie 4, Aachen, Germany (http:// www.linie4-architekten.de) helped to re®ne our work and evaluated it. Their support is gratefully acknowledged. We would also like to thank the reviewers for their valuable comments. References [1] R. Bayer, Document management as a database problem, in: U. Dayal, P.M.D. Gray, S. Nishio (Eds.), Proceedings of the 21st International Conference on Very Large Data Bases (VDLB'95), Zurich, Switzerland, 1995, pp. 7±10. [2] R. Bentley, W. Appelt, U. Busbach, E. Hinrichs, D. Kerr, K. Sikkel, J. Trevor, G. Woetzel, Basic support for cooperative work on the world wide web, International Journal on Human Computer Studies: Special Issue on Novel Applications of the WWW 45 (6) (1997) 827±846. [3] M. Bichler, A. Segev, Methodologies for the design of negotiation protocols on E-markets, Computer Networks, Special Issue on Electronic Business Systems 37 (2) (2001) 137±152, this issue. [4] T. Bray, J. Paoli, C.M. Sperberg-McQueen, Extensible Markup Language (XML) 1.0. http://www.w3.org/TR/ REC-xml, February 1998, W3C Recommendation. [5] S.P.M. Choi, J. Liu, S.-P. Chan, A genetic agent-based negotiation system, Computer Networks, Special Issue on Electronic Business Systems 37 (2) (2001) 195±204, this issue.

169

[6] J.L.G. Dietz, G.A.M. Widdershoven, Speech acts or communicative action?, in: L. Bannon, M. Robinson, K. Schmidt (Eds.), Proceedings of the 2nd European Conference on Computer-Supported Cooperative Work, ECSCW'91, Kluwer Academic Publishers, Dordrecht, 1991, pp. 235±248. [7] F. Dignum, J. Dietz (Eds.), Communication modeling ± The Language/Action Perspective, Proceedings of the 2nd International Workshop on Communication Modeling, Computing Science Reports 97-09, Eindhoven University of Technology, Department of Mathematics and Computing Science, Eindhoven, Netherlands, June 1997. [8] F. Dignum, R. Kuiper, Combining dynamic deontic logic and temporal logic for the speci®cation of deadlines, in: R. Sprague Jr. (Ed.), Proceedings of the 30th Hawaiian International Conference on Systems, Los Alamitos, 1997, IEEE Computer Society Press, Silver Spring, MD. [9] F. Dignum, H. Weigand, Communication and deontic logic, in: R.J. Wieringa, R. Feenstra (Eds.), Information Systems, Correctness and Reusability, Proceedings of the ISCORE-94 Workshop, Singapore, 1995, pp. 242±260. [10] T. Finin, R. Fritzson, D. McKay, R. McEntire, KQML as an agent communication language, in: Proceedings of the 3rd International Conference on Information and Knowledge Management (CIKM'94), ACM Press, New York, 1994. [11] F. Flores, J.J. Ludlow, Doing and speaking in the oce, in: G. Fick, R.H. Sprague Jr. (Eds.), Decision Support Systems: Issues and Challenges, Pergamon Press, New York, 1980, pp. 95±118. [12] G. Goldkuhl, M. Lind, U. Seigerroth (Eds.), Proceedings of the 3rd International Workshop on the Language Action Perspective on Communication Modelling, Department of Informatics, J onk oping International Business School, J onk oping, Sweden, June 1998. [13] G. Goldkuhl, M. Lind, U. Seigerroth (Eds.), Proceedings of the 4th International Workshop on the LanguageAction Perspective on Communication Modelling (LAP 99), Department of Informatics, J onk oping International Business School, J onk oping, 1999. [14] R. Grimm, P. Ochsenschlager, Binding telecooperation ± a formal model for electronic commerce, Computer Networks, Special Issue on Electronic Business Systems 37 (2) (2001) 171±193, this issue. [15] J. Gulbins, M. Seyfried, Strack-Zimmermann, Dokumenten-Management, Springer, Berlin, 1999. [16] J. Habermas, Theorie des kommunikativen Handelns, vol. 1: Handlungsrationalitat und gesellschaftliche Rationalisierung, Suhrkamp, Frankfurt am Main, 1981, Taschenbuchausgabe, 1995. [17] E. Van Heck, P. Ribbers, Experiences with electronic auctions in the Dutch ¯ower industry, in: Global Electronic Commerce: Theory and Case Studies, MIT Press, Cambridge, MA, 1999, pp. 355±366. [18] S. Klein, C. Loebbecke, Signaling and segmentation on electronic markets: innovative pricing strategies for improved resource allocation, in: Proceedings of the 6th

170

[19] [20]

[21] [22] [23]

[24]

[25]

[26]

[27]

[28]

M. Schoop, C. Quix / Computer Networks 37 (2001) 153±170 Research Symposium on Emerging Electronic Markets (RSEEM), M unster, Germany, 1999. K. Lyytinen, Implications of theories of language for information systems, MIS Quarterly, March 1985, pp. 61±74. K. Lyytinen, E. Lehtinen. On information modelling through illocutionary logic, in: H. Kangassalo (Ed.), Proceedings of the 3rd Scandinavian Research Seminar on Information Modelling and Data Base Management, Tampere, 1984, Acta Universitatis Tamperensis, Ser. B, vol. 22, pp. 35±115. MEMO-Consortium, MEMO home page, www.abnamro.com/memo/, 2000. J.J.C. Meyer, R.J. Wieringa (Eds.), Deontic Logic in Computer Science ± Normative System Speci®cation, Wiley, Chichester, 1993. C. Mohan, A database perspective on Lotus Domino/ Notes, in: Proceedings of the ACM SIGMOD International Conference on Management of Data, ACM Press, Philadelphia, PA, 1999, p. 507. C. Quix, M. Schoop, Facilitating business-to-business electronic commerce for small and medium-sized enterprises, in: Proceedings of the 1st International Conference on Electronic Commerce and Web Technologies, EC-Web 2000, Springer, Berlin, 2000, pp. 442±451. M. Schoop, Habermas and Searle in hospital: a description language for cooperative documentation systems in healthcare, in: F. Dignum, J. Dietz (Eds.), Communication Modeling ± The Language/Action Perspective; Proceedings of the 2nd International Workshop on Communication Modeling, Computing Science Reports 97-09, Eindhoven University of Technology, Netherlands, 1997, pp. 117±132. M. Schoop, Towards e€ective multidisciplinary communication: a language-action approach to cooperative documentation systems, PhD thesis, The University of Manchester, UK, 1998. M. Schoop, A theoretical framework for speech act based negotiation in electronic commerce, in: Proceedings of the 6th Research Symposium on Emerging Electronic Markets (RSEEM), M unster, Germany, 1999. M. Schoop, An empirical study of multidisciplinary communication in healthcare using a language-action perspective, in: G. Goldkuhl, M. Lind, U. Seigerroth (Eds.), Proceedings of the 4th International Workshop on

[29]

[30]

[31]

[32]

[33] [34] [35] [36]

[37] [38] [39]

the Language Action Perspective on Communication Modelling (LAP 99), Department of Informatics, J onk oping International Business School, J onk oping, 1999, pp. 59±72. M. Schoop, J. K oller, T. List, C. Quix, A three-phase model of electronic marketplaces for software components in chemical engineering, in: Proceedings of the 1st IFIP Conference on E-Commerce, E-Business, E-Government (I3E), Zurich, Switzerland, October 2001. M. Schoop, T. List, To monitor or not to monitor ± the role of trusted third parties in electronic marketplaces, in: Proceedings of the 5th International Tagung Wirtschaftsinformatik (WI-2001), Augsburg, Germany, September 2001. M. Schoop, C. Quix (Eds.), Proceedings of the 5th International Workshop on the Language-Action Perspective on Communication Modelling (LAP 2000), Aachener Informatik-Berichte 2000-6, RWTH Aachen, Aachen, Germany, 2000. M. Schoop, D.G. Wastell, E€ective multidisciplinary communication in healthcare: cooperative documentation systems, Methods of Information in Medicine 38 (4±5) (1999) 265±273. J.R. Searle, Speech Acts: An Essay in the Philosophy of Language, Cambridge University Press, Cambridge, 1969. J.R. Searle, D. Vanderveken, Foundations of Illocutionary Logic, Cambridge University Press, Cambridge, 1985. D. Selz, P. Schubert, Web assessment ± a model for the evaluation and the assessment of successful electronic commerce applications, Electronic Markets 3 (1997) 46±48. M. Str obel, Communication design for electronic negotiations on the basis of XML schema, in: Proceedings of the 10th World Wide Web Conference (WWW10), Hong Kong, ACM Press, 2001, pp. 9±20. L. Suchman, Do categories have politics? The language/ action perspective reconsidered, Computer Supported Cooperative Work (CSCW) 2 (3) (1994) 177±190. D. Suciu, Managing web data, in: Proceedings of ACM SIGMOD International Conference on Management of Data, ACM Press, Philadelphia, PA, 1999, p. 510. E.M. Verharen, A Language-Action Perspective on the Design of Cooperative Information Agents, PhD thesis, Katholieke Universiteit Brabant, Tilburg, The Netherlands, 1997.