HICSS'01: Broker's Lounge - An Environment for ... - Semantic Scholar

Proceedings of the 34th Hawaii International Conference on System Sciences - 2001

Broker's Lounge - an Environment for Multi-Dimensional User-Adaptive Knowledge Management Matthias Jarke, Roland Klemke, Achim Nick GMD - German National Research Centre for Information Technology GmbH FIT Institute for Applied Information Technology Schloß Birlinghoven, D-53754 Sankt Augustin, Germany [email protected], [email protected], [email protected]

Knowledge

n& tio sa ali ion xtu isat nte nal Co rso Pe

Ex Co plic nc atio ep n tua & lis ati on

three levels and their transitions are shown in fig. 1. We see information brokering as the value-adding process of mediation between information demands and

Information

&

Co n Ca cept teg ua or lisa isa tio tio n n

Abstract The Broker's Lounge is a shell for knowledge structuring and dynamic user interface generation which supports the personalisation of both information structures and user interfaces, emphasising options for context change and multi-dimensional constraint propagation. Two major applications have been developed so far: ELFI, an advisor that manages knowledge about research programs in Germany such that proposers can identify appropriate funding schemes; and MarketMonitor, a tool that helps companies monitor the web pages of competitors, suppliers, and customers for early detection of changes in the market situation.

Data

1. Introduction Knowledge management is facing two important trends. On the one hand, an ever increasing amount of information available through online resources leads to information overload and makes the selection of relevant and important information difficult. As an effect, the quantity of available information increases while the quality decreases. On the other hand, the availability of high quality information as a key factor for the success or failure of an organisation is more important than ever. This so-called information gap motivates the need for knowledge specialised for information retrieval and information quality assurance. Information brokering solutions may help to reduce the information gap. As a process-oriented subdomain of knowledge management, information brokering is concerned with data, information and knowledge. Following Alavi and Leidner [2], we define data as raw unstructured symbols (such as text and numbers). In information brokering, information is defined as processed, conceptualised and categorised data. Knowledge is information that is made actionable by being contextualised and personalised. The

Figure 1. Data, information, and knowledge information offers (fig. 2). Three roles participate in the information brokering process: the provider offers information, the consumer demands information, and the broker mediates between the other two. Several challenging problems have to be solved by the broker. Firstly, she has to be a kind of domain expert in her area of brokerage to be able to understand the domain complexity and the used vocabulary. Furthermore, she has to understand the (ambiguously or incompletely formulated) consumer’s need correctly, and map it to supplier terms (which may even differ for different providers) to retrieve relevant information. Therefore, she has to create a domain model as a view to the corresponding (implicit) provider domain models and also map clients’ information needs to this model (see fig. 2). Further problems arise from setting up, organising and maintaining the broker’s domain model and from evaluating providers and their offers (domain representation or maintenance tasks).

0-7695-0981-9/01 $10.00 (c) 2001 IEEE

1


The quality of the information brokering process depends to a great extent on the knowledge available to the broker. Knowledge about available sources, the domain, consumers and other brokers is needed. Source Broker‘s Domain Model

Provider‘s Domain Model View

Interest Model View

task and process-oriented approach to information brokering, building on experiences with two specific brokering systems which have been applied by some thousand users in research resp. industry. Our domain modelling approach, shown in section 4, is strongly motivated by the use of domain models during profiling. Broker’s Lounge is described in section 5. Section 6 discusses some extensions currently in progress while section 7 concludes the paper with a brief summary.

2. Related work

Domain representation/ maintenance tasks

Provider

Client-oriented tasks

Broker

Consumer

Figure 2. Roles, tasks, and domain models knowledge is created in the domain representation or maintenance processes and describes the quality of sources and how they can be accessed. Domain Knowledge is about the contents of the brokering domain and should reflect the provider’s understanding of the domain as well as the consumers’ perception of the domain. Consumer Knowledge is created in the consumeroriented process and describes the consumer and his concrete information need. Consumer knowledge has to mapped onto domain knowledge to fulfil the information need. To ensure an optimal service to consumers, they should be served by the best broker according to their information need. This assignment task depends on the availability of Expertise Knowledge about different brokers. Information technology can help to unburden brokers from routine tasks (e.g. check if a provider updated his web-site) in order to focus on intellectually more challenging or interesting tasks. However, until recently, developing information brokering systems had to be done on a case-by-case basis. Brokering methods and tools can be deployed in many different scenarios, with many different assignments of roles to tasks, different degrees of automation, etc. Based on experiences in developing and evaluation several different brokering systems, we have tried to obtain a systematic perspective on these tasks and their associated support demands. We then developed a brokerage development and management environment called the Broker’s Lounge in which a large variation of scenarios within the general framework of figures 1 and 2 can be quickly developed and efficiently and flexibly executed. The rest of this paper is organised as follows: after discussing related work in section 2, section 3 defines our

In this section, we review research from information brokering, knowledge management, and domain modelling that has influenced the problem analysis or solution strategy of our approach.

2.1. Information brokering A task analysis of telephone operators in [20] confirms that even such relatively mundane information brokers perform knowledge-intensive tasks and could thus benefit from knowledge management technologies. Our approach to task structuring was influenced by work in the European project COBRA [29], [15]. The literature shows informal descriptions and local tool support for many different information brokering scenarios, without a real integration. For example, [34] defines a pragmatic informal model of four roles in the case of organisation-internal information brokerage: users, domain experts, information experts and internal information brokers which confirms that human brokers improve the quality of delivered information drastically. A push approach to editor-based information brokering using a simple brokering model in [26] exemplifies several attempts to show the use of combining human and machine intelligence in knowledge management. The role of terminology management in information brokering is frequently stressed. For example, the GlOSSserver (Glossary-of-Servers-Server) [31] contains “summary information” (index of all keywords and their occurrence frequency) of multiple databases to address the problem of automatically selecting a database that is appropriate for a certain query. The architecture for network-based information brokering in [9] relies on modelling the relationships between domain models and source models. An application in the domain of health care is described in [10]. Both approaches use the Ontolingua system [11] for representation and modelling. Successful applications of domain and source modelling for the brokering of more structured information have also been reported for data integration in data warehouses and federated database systems [12], [17].

0-7695-0981-9/01 $10.00 (c) 2001 IEEE

2


2.2. Domain modelling A threefold ontology for organisational memories is motivated in [1]. It comprises an enterprise ontology modelling contextual information, a content ontology modelling domain contents, and an information ontology that models access to different information sources. In terms of information brokering tasks, the enterprise ontology guides contextualisation while the domain ontology supports conceptualisation. The information ontology is related to source evaluation. This approach lacks support for personalisation. An approach similar to Broker’s Lounge is described in [19]. The authors present an ontology-based domain modelling approach built on concepts and instances that allows attribute-, concept- and text-based queries. Their ontology allows inheritance and instance-of relations and is used to (manually) contextualise documents. In contrast to our brokering-focussed modelling approach, it lacks the separation of a concept level and a structuring category level, which we consider essential for efficient filtering of large amounts of information. As an example, we mention the problem of knowledge extraction from text via domain ontologies discussed in [8] and [4]. Three user groups are mentioned in [8]: journalists submit news stories to the system; knowledge editors annotate the stories according to the ontology; and consumers read the stories. Consumers can create interest profiles to filter new stories.

2.3. Knowledge management Our extensible toolkit architecture supports a processoriented view on knowledge management with strong emphasis on personalisation and contextualisation. We therefore review work related to these topics. A review of several knowledge management and corporate memory approaches is presented in [7]. Here, as in many other papers we find that contextualisation and personalisation in knowledge management are important but still open issues. According to [1], knowledge is personalised or subjective information related to facts, procedural concepts, interpretations, ideas, observations, and judgements. Consequently knowledge resides in the user, not in the collection of information. In contrast to papers such as [18], this demands that, to share (personalised, internalised) knowledge, it has to be communicated in an interpretable way. In [25], three forms of context are identified: external knowledge, contextual knowledge and procedural context. We build on these ideas by identifying information brokering processes that perform or support transformations between data, information, and knowledge. An information search and retrieval process model in [21] identifies context as an important aspect in information retrieval. Retrieval

processes are identified as asynchronous communications where the creation and retrieval contexts are different but important. A context-based model of communication is designed, not giving further details on context models. Knowledge management has been a long-standing topic in the knowledge engineering community [30], recently very much focused on classifying different kinds of formal ontologies: domain ontologies vs. generic ontologies valid across several domains vs. special-purpose application ontologies; and representational ontology defining an ontological framework. Our ontological work can be seen as belonging to generic application ontologies, where the application domain is information brokering, while the content domain is not predefined. This classification also has its counterpart in knowledge management systems architectures [3]. Vertical KM systems are developed for a specific domain, whereas horizontal KM systems such as Broker’s Lounge are conceived as frameworks and must be customised to a domain. The CommonKADS knowledge engineering methodology [28] suggests a layered expertise model for knowledge-based systems. The domain layer describes static knowledge needs, the inference layer describes the structure of inferences, while the task layer organises tasks into subtasks. Orthogonally, [32] provides a methodology for ontology building, proposing that ontology construction should start from basic level categories. Turning now to the process view of knowledge management and to the need for a wide range of task-role assignments, a characterisation of knowledge work using the well-known definitions of tame and wicked problems from the management literature can be found in [6]. Wicked problems must be addressed by a less structured more creative KM process. An example is the collaborative construction of concept indexes described in [22]. Here, documents are seen as a means of knowledge distribution and communication. Concept index and the documents together can be seen as a group memory (or collective memory), where concept index reflects a group specific view on a set of documents. This motivates a community brokering scenario where all members of a community can be providers, consumers, and brokers. In [33] corporate memories are defined and organised along two dimensions: active vs. passive collection of information and active vs. passive information distribution. These dimensions reflect from an information brokering point of view the possible task distributions among different stakeholders in the brokering processes. In [16] a 10-step model of organisational information processing consisting of recording, individual learning, information sharing, institutionalisation, action, feedback, repackaging and reproduction, communication and dissemination, and internal communication is defined. This model complements our information

0-7695-0981-9/01 $10.00 (c) 2001 IEEE

3


brokering models, as it models the process from an organisational rather than an individual point of view. The key idea of the process-oriented view on knowledge management in [13] is the explicit representation of mnemonic processes (processes to create, use and maintain knowledge) as business processes, following the underlying hypothesis that business processes involving people and technology are that part of an organisational memory that promises best utilisation of resources. Consequently, capturing and accessing operations for organisational memories should concentrate on these processes. The authors identify four basic knowledge agents (roles): knowledge creator, knowledge user, expert, knowledge administrator. These roles, derived from Takeuchi’s and Nonaka’s modes of knowledge conversion (socialisation, externalisation, combination, internalisation) are closely related to our role perception in information brokering processes

3. Process view of information brokering In this section, we analyse the tasks prevalent in information brokering and present a process framework in which different kinds of brokering scenarios can be described. We illustrate the usage of this framework by contrasting two real-world information brokering systems.

3.1. Task analysis Based on analysis work performed in several domains (see [15], [23], [29]) we identified a set of generic tasks important to information brokering processes. These tasks fall in roughly two groups: domain representation/ maintenance and consumer-oriented tasks. Domain representation/ maintenance tasks deal with capturing, organising, representing, and maintaining domain knowledge to prepare consumer-oriented tasks. They include source evaluation, conceptualisation, contextualisation, and categorisation. Consumer-oriented tasks are related to a specific consumer’s information need. Relying on the existence of represented domain knowledge, these tasks (subsumed under the label of personalisation) include request / assignment, profiling, querying, result selection, and delivery. We now describe these tasks in more detail. Of special interest for the rest of this paper are conceptualisation, contextualisation and personalisation. Source Evaluation. From the huge number of sources available a broker has to identify the ones delivering most promising results. She also has to find out about technical details of interaction with specific sources (e.g. how and where is information stored? How can I access it?). Conceptualisation. To organise, understand, and

evaluate incoming data, the broker has to find out what it is about. This implies the necessity to structure it along domain-dependent schemes, including the possibility to refine these. Information structured along those schemes has the advantage of being comparable and storable. Categorisation. To survey a domain, a specific classification scheme (category system) has to be applied to available information. Information (plain or conceptualised) that is categorised using such a scheme can be retrieved, filtered, grouped, and sorted. The applied classification system is meta information about the domain. Categorisation also comprises maintenance and administration of the category system. Contextualisation. Information does not offer a value in itself, it is only useful in appropriate contexts. Therefore it is necessary to annotate (enrich) information (which may be plain, conceptualised, or categorised) with appropriate contextual information (i.e. domain knowledge and situational information) in order to evaluate its relevance for a given domain or situation. Personalisation. The request /assignment task initiates a consumer-oriented personalisation process. By giving the request, the consumer outlines her information need. During the request task the broker tries to understand this need gathering additional information about the consumer. Intertwined with this is an assignment task leading to the selection of the most appropriate broker for a certain request. The selected broker starts an iteration of profiling, querying, result selection, and delivery, until the client’s information need is satisfied. Profiling is the task of expressing a consumer request using domain specific classification schemes and search expressions. Querying is the task of applying a profile to a selected set of domain-relevant sources resulting in a set of possible results. Result selection is the task of selecting the best results from the proposed set. Delivery is the task of forwarding the selected results to the consumer.

3.2. Brokering process models As a starting point for structuring the Broker’s Lounge, we embedded the individual tasks identified previously in a generalised brokering process model, called the information brokering knowledge cycle (see figure 3). Conceptualisation is performed on the basis of existing domain knowledge of the person performing this task, and optionally annotated documents, resulting in domain concepts. Contextualisation uses incoming documents and domain concepts to create annotated (or contextualised) documents. Personalisation is performed using either domain concepts or annotated documents (depending on the kind of brokered item) to select the most appropriate ones according to the consumer’s need.

0-7695-0981-9/01 $10.00 (c) 2001 IEEE

4


Conceptualisation Categorisation

Annotated Document

Domain Concepts

Personalisation

Contextualisation Documents

Figure 3. The information brokering knowledge cycle compare two rather different information brokering processes concerning their task distribution to different roles and the kind of information brokered.

Provider

Domain Concepts automatic


Domain Knowldege human

Contextualisation

automatic

Figure 4. The ELFI brokering process

human

Personalisation

Domain Concepts


Personalisation

Domain Concepts

Annotated Documents

Documents

Provider

Contextualisation

Domain Concepts

Consumer

Annotated Document

External Broker

Documents

3.2.3. Comparison. The ELFI and MarketMonitor brokering processes display some fundamental differences but also share common aspects in terms of the general cycle shown in fig. 3. Three main differences concern the brokered item, the task distribution on different roles, and the brokering process organisation (Table 1). (1) In ELFI, the brokered items are domain concepts while MarketMonitor brokers annotated documents. This difference stems from domain properties: in ELFI the types of concepts dealt with are well defined (namely

Internal Broker

3.2.1. ELFI. ELFI is an information brokering system in the area of research funding. Information providers are funding agencies offering information about their funding programs. Consumers are researchers who want to get their research funded without spending too much time on finding appropriate funding opportunities. About 1200 researchers in Germany are currently using the system. ELFI’s brokering process is in three stages (see fig. 4). Firstly, the external broker (the ELFI service provider) sets up the initial ELFI domain model, resulting in a set of domain concepts and classification terms. Secondly, automatic processes contextualise (annotate) documents gathered from information providers and a human broker conceptualises and categorises the contextualised documents in order to create new domain concepts. Thirdly, the internal broker (i.e. a funding agent at the

3.2.2. MarketMonitor. MarketMonitor, a brokering system developed with and marketed by humanIT GmbH, offers a semi-automatic solution that monitors market and competition information from different online information sources (see fig. 5). Providers are news services and competitors, represented through their online information services. Consumers are decision-makers of the organisation that runs a MarketMonitor service. The broker is an internal one: she is a member of the organisation and uses her domain knowledge to specify the organisation view by domain concepts and categories. In an automatic process, documents are gathered from the provider sites and contextualised along the domain concepts resulting in annotated documents. Personalization step is done by the consumer himself who specifies his interest in terms of queries and retrieves the most appropriate annotated documents from the repository.

Domain Knowldege

Annotated Documents

Consumer

Domain Knowldege

researchers university) personalises the conceptualised information to the researcher’s need by specifying interest profiles which filter the most appropriate domain concepts out of the set of available concepts.

Internal Broker

This general framework allows for a wide variety of, actual brokering scenarios. In the sequel, we use it to

Figure 5. The MarketMonitor brokering process funding programs and corresponding contact information) while in MarketMonitor, concepts are a helpful means to state what news are about; the brokered information is the news about concepts instead of new concepts.

0-7695-0981-9/01 $10.00 (c) 2001 IEEE

5


(2) In ELFI we have to cope with a twofold brokering process, where brokering tasks are distributed on external brokers (dealing with contextualisation and conceptualisation) and internal brokers (dealing with personalisation). In MarketMonitor, only an internal broker exists who is responsible for conceptualisation and contextualisation. Personalisation is done by the consumer herself. The reason for this difference is manyfold: The group of ELFI consumers, interested in the same kind of information, is rather big (potentially all German scientists). This requires a central institution offering high quality conceptualised and contextualised information. Personalisation in turn requires good knowledge of the consumer and thus is not performed by the central service provider. In MarketMonitor, the number of consumers is relatively small and the domain knowledge is very specific to the organisation. These conditions do not justify an external broker. Furthermore, as the interest in news is rather short term, personalisation is done by the consumer in an ad hoc manner, while ELFI researchers have a long term interest in their specific area of research, resulting in long term profiles maintained by the internal broker. (3) ELFI offers a continuous process of contextualisation, conceptualisation and personalisation. MarketMonitor has two orthogonal processes, one dealing with conceptualisation, one with contextualisation and personalisation. The reasons for this difference are as follows. In MarketMonitor, the organisation (acting as internal broker) conceptualises its own view of the domain. This view remains relatively stable over a period Domain structuring

ConceptType

Category

Concept

Defines type

CategoryType

Is a

Type of

Related to

Instance of

Classifies

domain view which is continuously expanded through incoming documents. For the consumers it is interesting to monitor exactly these changes in the conceptualised information.

4. Domain modelling Domain modelling is an essential part of successful information brokering. The domain model represents the broker’s domain view and is used for consumer-oriented profiling. Typically, domain experts without technical skills work with the domain model. Therefore editing the domain model should be intuitive and the domain experts should not need skills in formal languages to extend it. Each information broker deals with a specialised information domain – concepts such as funding programs (scholarships, research prizes,..), funding agencies, and contact persons of funding programs. Interest profiles can be formulated extensionally by listing interesting concepts, or intentionally by describing desired characteristics. Intentional profile definition requires a separate structuring level, allowing to group concepts (independent of their actual existence). An intentional profile is more valuable, as it can be applied to new concepts in the domain model, whilst an extensional model can only be applied to known concepts. Mostly, category systems (e.g. lines of business in yellow pages) are used as structuring level to define intentional profiles. To ensure that category-based intentional profiles are useful, all concepts have to categorised. Typically, categories are organised in hierarchies and can be used for filtering (e.g. all funding opportunities for natural sciences without chemistry). Many category systems suffer from the need to express multiple facets in a single hierarchy. We use multihierarchies instead, with a different category type for each facet. In ELFI we have two category types (research topic and funding type) classifying funding programs. As information brokers usually inform about classes of related concepts, we introduced concept types, too. In ELFI, e.g., funding agencies are of interest (there the money comes from!) besides funding programs. Based on these ideas we designed the basic object structure for the domain model as depicted in fig. 6. Relations between concepts and categories are defined on

Figure 6. Basic classes for domain models of time. Members of the Table 1. Dimensions of information brokering organisation want to map news from ELFI MarketMonitor the external world onto this view. Brokered Item Domain Concepts Annotated Documents This interest leads to the second Task External and internal brokering with Internal brokering with external process, that contextualises and distribution external sources sources personalises external information Main Two orthogonal processes: preparation One continuous process through along the domain concepts. In ELFI Brokering contextualisation, conceptualisation, by conceptualisation, brokering by contextualisation and personalisation and personalisation the external broker defines an initial Processes

0-7695-0981-9/01 $10.00 (c) 2001 IEEE

6


Funding Agency

BMBF

ConceptType

Research Topic

Funding Programm

Contact Person

Region

Engineering Sciences

Multimedia Book

Dr. Peter Muster

Europe

Computer Science Is a

Type of

Germany Related to

Instance of

Instance-Level Type-Level

CategoryType

Meta-Level

the type level, concrete instances and their attribute values are supplied on the instance level (see fig. 7). The domain model is an ontology [30] as it is a formal (i.e. system readable) and shared description of the modelled domain that reflects the broker’s view to the domain, modelled on behalf of her clients (consumers). Broker’s Lounge supports instance-of-, is-a- and part-ofrelations and simple associations. Profiles are used to select interesting concepts out of all available ones. Filtering can be done using the four building blocks (categories, category types, concepts and concept types) of domain models, which we illustrate along the ELFI domain: Concept type filters exclude certain concept types (and their concepts). E.g. an ELFI user may be interested in funding programs but not funding agencies. Category type filters can be used, if there is more than one category type for a certain concept type. Funding programs have several category types (research topic, interdisciplinary topic, funding type,...). If a user is

Classifies

Figure 7. A part of the ELFI ontology interested in all funding types, but only in funding for special research topics, a category type filter for research topics is needed, but no filter for funding type. Category filters are based on category hierarchies where selected categories determine resulting concepts. The funding program “Multimedia Book” (fig. 7) fits the research topic filter, if “Computer Science” is selected. Concept filters can simply list concepts or they reflect certain attributes or concept relations. Filtering funding programs funded by the Ministry for education and science (BMBF), is an example of filtering by relation. Different filters can be combined to model consumer interests but usually category filters play a major role.

5. Broker’s Lounge Broker’s Lounge is a knowledge management environment aimed to support the complete process

sketched above. It covers a variety of information brokering scenarios concerning task and role distribution among participants in the scenario. Furthermore, it is independent of the content domain. Broker’s Lounge supports domain experts to set up domain specific knowledge management solutions in a short period of time. We focused on the development of intuitive user interfaces supporting the stepwise development of domain models with no need of technical knowledge. We offer an integrated solution that is configurable to support a variety of brokering scenarios along several dimensions (e.g. task distribution on roles, brokered item, process design, compare Table 1). Using fig. 8, we will now discuss, how the different brokering tasks are realised in Broker's Lounge. The notion of tasks and roles allows the configuration of Broker’s Lounge for different brokering scenarios. Each user of the system is permitted a set of tasks. This flexible assignment may distribute tasks to the different stakeholders or summarise several tasks in one person. It also influences possible paths of information flow. A user who is assigned to the provider-oriented source evaluation task, interacts with the source administration user interface. This interface offers access to the source models which configure the system’s access to different online sources. The user can register or remove sources or change access profiles (stating access frequency and policy, or weighting sources to represent their evaluation). The contextualisation task is performed by software robots, configured by the source model. These robots gather documents from online sources. The documents are parsed, using the domain model as input. The matching Source Evaluation

Personalisation

Source Model

Robots

Profile Model Source Admin

Profile Browser

Tasks

Roles

Doc Viewer

Ontology Admin

Parser Contextualisation

Domain Model Document Index Conceptualisation Categorisation

Figure 8. System architecture algorithm scans the documents for occurrences of domain concepts (possibly with multiple synonyms) calculating a domain score for each. The parsing result, a document annotated with concepts, is stored in the document index. The conceptualisation / categorisation task is supported through the ontology administration interface in

0-7695-0981-9/01 $10.00 (c) 2001 IEEE

7


combination with the document viewer. The document viewer allows to browse through contextualised documents, querying them along various dimensions (actuality, matched concepts, matched categories, kind of source). The ontology administration interface is used to browse through collections of concepts, add new concepts and categories and edit existing ones (see figure 9). The personalisation task is supported through the profile browser. Depending on the nature of the brokered item (e.g. domain concept vs. annotated document), it is either a personalised version of the document viewer, or a personalised version of the ontology admin. These personalised components are extended with profiling mechanisms, allowing to filter the available information along concept types, category types, concepts, and categories. We illustrate their use along fig. 10: a concept type filter describes the kind of information somebody is interested in. While Achim and Roland are interested in funding programs, funding agencies and contact persons, Matthias is interested in funding programs and funding agencies. A category type filter describes which (category) dimensions are used for filtering. Roland uses all dimensions, while the others only use some category types. A Category filter (represented by the hooks) is used to filter a special category

email (delivery task).

Figure 10. Profile browser

6. Extensions Our experiences shows that further research is needed in two fields: context modelling and interest modelling. 6.1. Comprehensive context modelling

Figure 9. Ontology administration dimension. Achim is e.g. interested in computer science and medicine in the research topic dimension. Concept filters are organised along concept properties or relations (e.g. an actuality filter). From a system point of view, actuality measures the last concept change, while to the user, actuality is relative to the dialog history. In our example Achim and Roland use an actuality filter (depicted by the small diary icon). Complex filters can be created by combining filters. Filtering results are displayed on the right hand side, where filter changes are reflected in changed result tables immediately (querying task). Concepts can be collected to a dossier (result selection task), which can be delivered by

In a literature study [14], we found many approaches regarding context as an important issue but formalisations of contextual information are lacking. Instead, every approach looks at context in a specific sense focussing on only some contextual aspects. Especially in internal brokering scenarios such as organisational memories, much context information is available. Generally, an organisational memory captures relevant organisational information (such as documents, domain models, process models and organisational models) in order to distribute it to organisational members who need it for their task at hand. Thus, the organisational memory serves as an information broker internal to the organisation, where providers and consumers are the same organisational members. Usually, members of a single organisation share a common range of working contexts within which information production and consumption occur. This drives our vision towards context-based organisational memories. The motivating question behind further research concerning context is: If we know about the context in which some information has been created or used and the context in which some person currently is, how can we provide relevant information to that person?

0-7695-0981-9/01 $10.00 (c) 2001 IEEE

8


In a general sense, context is any information that can be used to characterise the situation of an entity. Based on our literature study we defined a context typology that comprises four basic contextual dimensions (see fig. 11). To make use of contextual information within an organisation we have to meet two important requirements: contextual information has to be gathered in a semiautomatic way (1) and contextual information has to be comparable (2). Requirement 1 demands combining available organisational information sources to create context models. Such an approach could for instance be based on Context Toolkit [27]. Requirement 2 needs complex similarity measures for context models measuring how similar different situations are. Process (e.g. Workflow) Organisational

Structure (e.g. Enterprise Ontolgy) Domain Ontology

Domain/Content based

Knowledge Profiles

Context User Profiles / User Models Personal

Interest Profiles Location

Physical Time

Figure 11. Context typology

6.2. Adaptive interest modelling User modelling distinguishes between adaptable and adaptive systems. Adaptable systems are user-adapted, while adaptive systems are adapted automatically. Broker’s Lounge is an adaptable system but especially the profiling task is a candidate for adaptive features, as a profile reflects user interests and determines retrieved results. An adaptive component could assist the user in profiling by pointing to categories, that should be added to or removed from a given profile whenever a changed interest is observed. Also, the system can highlight relevant information items or items can be ordered by relevance. It is important that users control their profile, as (automated) changes can result in a loss of information. A difficult issue is the acquisition of knowledge about the user. Acquisition methods [5] can be divided in direct methods, where users must actively feedback or fill out questionnaires and indirect methods, where the system analyses its usage. Indirect methods are preferable, as they do not disturb the user’s actual work. As the domain ontology is a graph (see Fig 7) with several vertex types (reflecting relation types), the system

may observe the user’s path through the domain model. These observations are mapped against interest profiles. E.g., if an ELFI-user requests information about several funding programs for mathematics, he is probably interested in mathematics. As physics is closely related to mathematics (in the domain model) the user may be interested in physics, too. First experiences in interest modelling with Broker’s Lounge [24] are encouraging.

7. Summary We presented a process-oriented approach to the design and operation of information brokering solutions in knowledge management which has been implemented in the Broker’s Lounge environment. Broker’s Lounge offers a type-based multi-dimensional ontology. The domain independent ontology creation and maintenance possibilities in combination with filtering approaches based on personal interest profiles supports the personalisation of information structures. Streamlined with that, task and role specific interfaces, profile dependent ontology presentation, and ontology dependent query interfaces are the basis for personalised user interface generation. By combining different filtering approaches, multi-dimensional constraints are created and used to reduce information spaces. The integrated approach, supporting all different tasks and roles using one environment, offers the opportunity to reuse the system in different brokering contexts. Even though research concerning a fully structured approach to context modelling and a more automated adaptive interest modelling is still ongoing, our experiences from two real-world information brokering projects (ELFI and MarketMonitor), where we successfully used the Broker’s Lounge environment to create customised brokering solutions, are encouraging.

References [1] A. Abecker, A. Bernardi, K. Hinkelmann, O. Kühn, M. Sintek. “Toward a Technology for Organisational Memories“, IEEE Intelligent Systems & Their Applications, May/June 1998. [2] M. Alavi, D. Leidner. “Knowledge Management Systems: Emerging Views and Practices from the Field”, Proc. of 32nd Hawaii Int’l Conf. on System Sciences, Hawaii, 1999. [3] V. R. Benjamins, D. Fensel, and A. G. Pérez. “Knowledge management through ontologies”, 2nd Int’l Conf. on Practical Aspects of Knowledge Management (PAKM98), Basel, 1998. [4] B. Biébow and S. Szulman. “Terminae: A linguistic-based tool for the building of a domain ontology”, Knowledge Acquisition, Modeling and Management - 11th Europ. Workshop, EKAW'99 (LNAI 1621), pp. 49-66. Springer, 1999. [5] Peter Brusilovsky: “Methods and techniques of adaptive hypermedia”, in: User Modeling and User-Adapted Interaction, 6(2-3):87-129, 1996.

0-7695-0981-9/01 $10.00 (c) 2001 IEEE

9


[6] S. Buckingham Shum. “Negotiating the Construction and Reconstruction of Organisational Memories”, Universal Computer Science, 3(8), Springer 1997 [7] R. Dieng, O. Corby, A. Giboin, M. Ribière: “Methods and Tools for Corporate Knowledge Management”, Proc. 11th Workshop on Knowledge Acquisition, Modelling and Management KAW’98, Banff, 1998 [8] J. Domingue and E. Motta. “A knowledge-based news server supporting ontology-driven story enrichment and knowledge retrieval”, Proc. 11th European Workshop, EKAW'99 (LNAI 1621), pp. 103-120. Springer, 1999. [9] R. Fikes, R. Engelmore, A. Farquhar, & W. Pratt. “Network-based Information Brokers”. Technical Report KSL95-13, Stanford University, January, 1995. [10] J. H. Gennari, D. E. Oliver, W. Pratt, J. Rice, M. A. Musen. “A Web-Based Architecture for a Medical Vocabulary Server”. Technical Report KSL-95-41, Stanford University, 1995. [11] T. R. Gruber. “A Translation Approach to Portable Ontology Specifications”. Technical Report KSL 92-71, Stanford University, 1993. [12] M. Jarke, J. Köller, T. List: “The challenge of process data warehousing”. Proc. 25th Intl. Conf. Very Large Data Bases (VLDB 2000), Cairo, September 2000. [13] R. Klamma, S. Schlaphof (2000). “Rapid Knowledge Deployment in an Organizational-Memory-Based Workflow Environment”, 8th European Conf. On Information Systems, ECIS 2000, Vienna, July 2000. [14] R. Klemke. "Context Framework - an Open Approach to Enhance Organisational Memory Systems with Context Modelling Techniques", accepted for: PAKM2000: 3rd Int’l Conf. on Practical Aspects of Knowledge Management, 30.-31. October 2000, Basel, Switzerland. [15] R. Klemke, J. Koenemann. „Supporting Information Brokers with an Organisational Memory“, XPS-99, 5th German Conf. on Knowledge-Based Systems, Workshop on Knowledge Management, Organizational Memory and Reuse, Internal Report, Würzburg University, March 1999 [16] F. Lehner, R. Maier, O. Klosa: "Organisational Memory Systems: Application of Advanced Database & Network Technologies in Organisations", 2nd Int. Conf. On Practical Aspects of Knowledge Management (PAKM98), Basel, 1998. [17] A. Levy, D. Srivastava, T. Kirk: “Data model and query evaluation in global information systems. “ Journal of Intelligent Information Systems 5, 2, 1995, pp. 121-143. [18] S. Mahé, C. Rieu: "A Pull Approach to Knowledge Management", Proc. 2nd Int. Conf. On Practical Aspects of Knowledge Management (PAKM98), Basel, 1998. [19] M. Mach, T. Sabol, J. Paralic, R. Kende. “Knowledge Modelling in Support of Knowledge Management”, Proc. Conf. on Research Information Systems, CRIS 2000. [20] M. J. Muller, R. Carr, C. Ashworth, B. Diekmann, C. Wharton, C. Eickstaedt, J. Clonts. “Telephone Operators as Knowledge Workers: Consultants Who Meet Customer Needs”, Int’l Conf. on Computer Human Interaction, ACM, 1995. [21] L. D. Murphy. “Information Product Evaluation as Asynchronous Communication in Context: A Model for Organizational Research”, Proc. 1st ACM Int’l Conf. on Digital Libraries, 1996. [22] K. Nakata, A. Voss, M. Juhnke, T. Kreifelts. "Concept

Index: Social Knowledge Construction from Documents", ERCIM News Nr. 35 - October 1998. [23] A. Nick, J. Koenemann, E. Schalück: “ELFI: Information brokering for the domain of research funding”, Computer Networks and ISDN Systems, 30: 1491-1500, 1998. [24] W. Pohl, A. Nick: “Machine Learning and Knowledge Representation in the LaboUr Approach to User Modelling”, 7th Int’l Conf. on User Modelling, Canada, 1999 [25] J.-Ch. Pomerol, P. Brézillon: "Dynamics between contextual knowledge and proceduralized knowledge", 2nd Int’l Conf. Modeling and Using Context (CONTEXT'99), Trento, Italy, 1999, LNAI 1688, Springer. [26] A. Rudström, A. Waern, K. Höök. “Interactive adaptation of Intranet newsletters”, Proc. workshop on Adaptive Systems and User Modelling on the World Wide Web, 6th Int’l Conf. on User Modelling, Chia Laguna, Sardinia, 1997. [27] D. Salber, A. K. Dey, G. D. Abowd. “The Context Toolkit: Aiding the Development of Context-Enabled Applications”, CHI 99, May 1999 [28] A. Th. Schreiber, B. J. Wielinga, R. de Hoog, H. Akkermans, W. van de Velde: “CommonKADS: A Comprehensive Methodology for KBS Development”, IEEE Expert, 28-37, December 1995 [29] R. M. Strens, M. Martin, J. E. Dobson, S. Plagemann. „Business and Market Models of Brokerage in Network-Based Commerce”, Proc. 5th Int. Conf. on Intelligence in Services and Networks (IS&N’98), Antwerp, Belgium. LNCS, Springer, 1998. [30] R. Studer, V.R. Benjamins, D. Fensel: “Knowledge Engineering - Principles and Methods”, Data & Knowledge Engineering 25 (1-2), March 1998. [31] A. Tomasic, L. Gravano, C. Lue, P. Schwarz, L. Haas. “Data Structures for Efficient Broker Implementation”, ACM Trans. on Information Systems, 15(3), July 1997 [32] M. Uschold and M. King. “Towards a methodology for building ontologies“, Workshop on Basic Ontological Issues in Knowledge Sharing, IJCAI-95, 1995. [33] G. van Heijst, R. van der Spek, E. Kruizinga. “Organising Corporate Memories”, Proc. 10th Workshop Knowledge Acquisition for Knowledge-Based Systems, 1996. [34] C. M. Vishik: "Internal Information Brokering and Patterns of Usage on Corporate Intranets", GROUP 97, Phoenix, Arizona, 1997

0-7695-0981-9/01 $10.00 (c) 2001 IEEE

10