Improving knowledge worker productivity - the ACTIVE integrated approach

Service oriented infrastructure (Part 2)   » Log in

Improving knowledge worker productivity – the ACTIVE integrated approach

P Warren, N Kings, I Thurlow and J Davies (BT), T Bürger and E Simperl (STI Innsbruck), C Ruiz and JM Gòmez-Pèrez (Intelligent Software Components), V Ermolayev (Cadence Design Systems GmbH), R Ghani (Accenture Technology Labs), M Tilly (European Microsoft Innovation Center), T Bösser (kea pro), and A Imtiaz (Forschungsinstitut für Rationalisierung)

E-mail a friend

E-mail the author

Thank you!

Knowledge workers are central to an organisation's success, yet the tools they must use often stand in the way of maximising their productivity. The research described here, which is being undertaken as a part of ACTIVE, an EU-funded collaborative research project that will run until February 2011, seeks to remedy some of the defects of current knowledge worker tools. It addresses the need for greater knowledge worker productivity by providing more effective and efficient tools. Three themes drive our research. First, we are concerned with easier sharing of information through a combination of a formal approach based on ontologies and informal techniques such as user tagging and the use of wikis. Second, we are prioritising the delivery of information through an understanding of the user's current task context. And finally, we are helping users to share and reuse informal processes, including by learning those processes from the user's behaviour. The results are relevant to all knowledge work, but are being validated in the domains of telecommunications, consultancy and engineering. Validation is done at the system, user and organisational levels.

The work described in this paper is aimed at materially improving the productivity of knowledge workers in all sectors of the economy. We are concerned both with knowledge workers who work alone and with those who work in teams. Our goal is to improve the performance of the tools they use, offering them a more efficient and effective interface to the underlying knowledge they use. The work is being undertaken in an EU collaborative project, ACTIVE, coordinated by BT, details of which are available online [1].The project will run for three years until February 2011.

The term 'knowledge worker' is generally attributed to Peter Drucker (e.g., in [2]). More recently, Thomas Davenport has defined knowledge workers as people who "think for a living" and discusses who they are, what they do, and how their number is growing [3]. For our purposes we accept as a given that, for an increasing proportion of people in the world economy, work is to a large extent mental rather than physical. To significantly increase economic productivity, it is necessary, therefore, to increase the productivity of this knowledge-based and knowledge-driven work.

While knowledge workers appear often to be working alone - for example, when writing reports - they are often critically-dependent on colleagues. The documents they create are often multi-authored and in any case they frequently need the knowledge of others to assist them in achieving their own goals. Therefore, knowledge sharing is critically important to them. Examples of the kinds of knowledge that need to be shared are the results of scientific experiments, engineering designs which can be reused by others and the text of customer proposals that can in part be reused to develop business with other customers. We also know that knowledge workers are dependent on a variety of software tools running on computers or other electronic devices. In fact, the tools knowledge workers use can generally be classified into two groups:

- those that are generic and used by all knowledge workers, such as word-processors and email; and
- those that are domain-specific, such as engineering design tools used by specialist engineers. Our work considers both sorts of tools.

A significant role of these tools is to deliver information¹ to their users. However, knowledge workers can frequently be overwhelmed by information but still lack that which is germane to their current task. We believe that the concept of task context is central to improving this situation. Knowledge workers need easy access to information which is relevant to their current task. It is vital to their productivity that information currently relevant is given prominence over that which is not.

When knowledge workers need to switch from one task to another, they therefore need an environment in which context switching is quick and simple, and in which the previous context is saved so that it can be restored instantly at a later stage. All knowledge workers engage in processes. Some are formal business processes, defined by the organisations in which the people work. In many cases, workers are not allowed to deviate from them. Languages such as the Web Services Business Process Execution Language (WS-BPEL) [4] for process execution and BPMN (Business Process Modelling Notation) [5] for process description can be used to describe them. However, we observe that knowledge workers often create their own informal processes. Examples are:

- A salesperson obtaining information on a customer before a sales visit. Typically, such a process will involve visits to a number of public and corporate websites, use of a number of corporate IT systems and interactions with a variety of colleagues.
- A corporate employee navigating a corporate system, for example to record customer information, possibly drawing in information from other systems and/or web-sites.
- Skilled engineers using a sequence of technological operations to progress a design. The operations involved will depend on the specific character of the design artefact. The way in which they are used - for example, the ordering of the operations - may depend on the level of skill of the engineer.

Such processes are rarely documented or, if they are, only very informally. This hinders their reuse even by their creators, and is a significant barrier to the sharing of methods between colleagues. We need, therefore, to provide assistance to knowledge workers in creating, reusing, sharing and also improving on these knowledge processes.

---

¹ The observant reader may have already noted that we sometimes use the word 'information' and sometimes the word 'knowledge'. There has been much debate over the difference in meaning between these two words. However, our goals are practical, and we do not find such debates helpful to these goals. In our paper, 'knowledge' is used in phrases where this usage is common practice - for example, 'knowledge worker', 'knowledge management' and 'knowledge description'. The term 'information' is used in less specific contexts. Generally we use whichever word seems the more natural English usage.

The discussion in section 1 highlights three key research challenges:

- improving knowledge sharing;
- managing information by taking account of the user's task context; and
- supporting knowledge workers in their use of informal processes.

2.1 Knowledge sharing
Ways of sharing knowledge effectively have been widely studied, both from a technological and an organisational viewpoint. The challenge is to ensure that the knowledge that exists within an organisation is available to all who need it. For some global organisations, this means that knowledge created on one continent needs to be made available to workers in others. Sometimes, the creators and users of the knowledge will not know, or even be aware of, each other. The innovation which we bring to this problem is to combine two parallel approaches to knowledge representation: formal and informal.

Since the beginning of this decade, considerable progress has been made in the development of formal techniques for describing and reasoning about knowledge. These advances are referred to broadly as 'semantic technologies' and are characterised by the use of ontologies to represent knowledge. (For an introduction to the use of ontologies in knowledge management see [6].)

In parallel, informal tags have been used to describe documents and media objects on the web, initially in the hobbyist and consumer worlds. Such tags are said to constitute 'folksonomies'. Flickr [7], the web-site for sharing photos, is an example of a site that uses them. Recently, organisations have begun to use similar techniques to expedite knowledge sharing. Ontologies and folksonomies offer contrasting advantages and disadvantages. Ontologies enable knowledge to be described in sufficient detail for machines to reason based on the descriptions they provide. However, the development and maintenance costs - for example, the cost of creating and updating the associated metadata - are relatively high. Folksonomies are easy to use but lack the descriptive richness required for automated reasoning. To obtain as far as possible the benefits of both techniques and minimise the disadvantages, we are combining the two approaches. For example, from observation of how tags are used, we hope to deduce structure that was not explicitly defined by users, and hence to define and refine the ontologies involved.

        Figure 1. A possible ACTIVE user interface

The use of folksonomies forms part of a phenomenon known as 'Web 2.0', characterised by the ability of individual users to create their own content. Also typical of Web2.0 is the use of wikis for knowledge sharing. Recent work has begun to marry the formal ontology-based approach with the informal approach of the wiki, resulting in the semantic wiki - for example, as discussed in [8]. Within the ACTIVE project, we are taking this approach further while at the same time using it as a test bed for the development of OWL2², the next version of the standardised language used to represent ontologies and ontology-based information.

2.2 Task context
The exploitation of context has already been the subject of significant research (for example, see [9]). Context, though, means different things to different authors. To some, the concept is a combination of the environment in which the person is working (for example, in or away from the office) and the tools he or she is using (such as a laptop or PDA). Others view context as being more about the specific task an individual is performing. Our interest is with this latter meaning and, to make this clear, we use the phrase 'task context'. The challenge here is to learn task context from the user's behaviour (e.g., web-browsing and email usage) and to prioritise information delivery according to the context that is deduced.

Our contention, borne out by discussions with typical users of information systems prior to the beginning of the project, is that, at any given time, users want easy access to information that is relevant to their current task context. For someone working with customers, the current context is likely to equate to the particular customer currently being considered. For a lawyer, the current context is likely to be the current case being considered. Figure 1 shows a snapshot of a possible ACTIVE user interface, illustrating how the context manager displays the list of possible contexts, indicating which is currently active.

At the bottom right is the context manager. This displays the user's five contexts. Four of these relate to customers. The fifth ('design work') is the user's current context and is generic; this context includes the generic design information which the user needs for many customers. On the left, the user is opening a file in the word processor. The four files shown are the most recently accessed files relating to the current context. Below this, the user is able to access files in other contexts. At the top right the user is accessing a wiki page, also relating to the current context.

To change context, the user simply clicks on a different context within the context manager. We are developing automatic techniques to enable the system itself to detect that the user is working on a different task, and suggest that the context should be changed. The user is, of course, always in control.

---

² OWL (http://www.w3.org/2004/OWL/) is a standard web ontology language created by the W3C. OWL2 is now being developed to extend the original language with a "small but useful set of features that have been requested by users". For details, see http://www.w3.org/TR/owl2-profiles/.

2.3 Informal processes
To date, the majority of research relating to processes has had to do with formal business processes. There has been some discussion of informal processes, often under names such as 'artful business processes', and a recognition that current IT systems do not cater well for such processes [10]. Our challenge is to enable informal processes to be created and edited by users, and also to create process descriptions automatically for users. To do this, we must (automatically) learn from a user, or users, the behaviours that constitute commonly-used processes.

Informal processes are carried out by knowledge workers with their skills, experience and knowledge - often to perform difficult tasks which require complex, informal decisions among multiple possible strategies to fulfil specific goals. In contrast to business processes that are formal and standardised, informal processes are often not even written down, let alone defined formally, and can vary from person to person even when those involved are pursuing the same objective. Knowledge workers create informal processes 'on the fly' to cope with many of the situations which arise in their daily work. While informal processes are frequently repeated, because they are not written down, they are not exactly reproducible, even by their originators, nor can they be easily shared.

Examples of informal processes can be found everywhere in the knowledge worker's life - for example, 'marketing a new product', 'resource planning' and 'proposal writing'. These examples take a relatively long time to complete, but informal processes can also be quite short - for example, a set of actions to update customer information on a database that is repeated regularly. Some can even include components that are quite formal - 'marketing a new product' might involve the use of formal processes when it comes to obtaining any corporate signoffs that are required, for instance. Often, as part of the process, the knowledge worker connects to other people - for example, to obtain relevant information.

Through ethnographic studies, the authors of [10] have identified that users are forced by today's systems to act as integrators of information across the numerous systems which they have to use, frequently cutting and pasting across systems that cannot easily share data let alone the semantics that apply to it. Such operations themselves form a kind of informal process - one that is relatively short in duration but frequently repeated. Usually, there are many possible ways to achieve the process objectives or to reach a certain goal. The knowledge worker has to make complex decisions to optimise his or her process and to reach the goal. Ideally, informal processes should improve through review and through an amalgam of colleagues' ideas. In practice, the fact that they are rarely written down is an impediment to this improvement.

2.4 The ACTIVE Knowledge Workspace
These three research challenges cannot be seen in isolation. Each has a bearing on the others. Knowledge sharing needs to be guided by the context in which an individual is working, and context itself is shareable. Knowledge sharing also needs to be guided by the particular processes the user is navigating, and processes can certainly be shared. Finally, task context helps to guide users through processes and the way in which knowledge processes are used can be a clue to task context.

To enable the functionality developed through the ACTIVE project to be fully integrated, an overall architecture has been designed that integrates individual components. Known as the ACTIVE Knowledge Workspace (AKWS), the architecture comprises a set of cooperative software services. In ACTIVE, we are not reproducing the common user applications. ACTIVE users will continue to use the applications with which they are familiar. This applies both to generic applications, such as word-processing and email, and to the specialist applications used in the project's case studies - for example, the engineering applications used in the Cadence case study (see section 6.3). The AKWS will interact with these applications so as to offer the benefits of ACTIVE to users through applications with which they are already familiar.

3.1 Knowledge sharing
As we have already observed, improving knowledge sharing within the organisation has been the subject of investigation for many years, both from the technological and organisational standpoints. A common solution is the deployment of some sort of taxonomic system. This requires the establishment of an agreed hierarchical taxonomy, and also requires that those using the system, whether to deposit or retrieve knowledge, must be familiar with the taxonomy. The recent enthusiasm for social tagging in the consumer domain has prompted the question whether this would be an appropriate and more accessible alternative within the organisation. It has also raised the question whether a synergy of the taxonomic and folksonomic approaches is feasible. Both questions are discussed in [11], where a system which attempts such a synergy is described - one that uses a thesaurus to suggest terms that match a tag entered by a user. Previous work by BT, one of the partners in ACTIVE, took a related approach combining user tagging and an ontology-based approach to the classification of information [12].

The use of wikis for knowledge sharing within organisations is now widespread. The association of some semantics with the hyperlinks within wikis has led to the semantic wiki. As an example, a link between a wiki page about London and a page about the UK could be annotated with the informal semantic 'is capital of'. This has the advantage of making richer querying and retrieval of information possible. The Karlsruhe Institute of Technology, another of the partners in ACTIVE, was one of the pioneers of the semantic wiki [8] and is taking this work forward as a part of the project.

3.2 Managing information by context
A number of research projects and commercial tools are under development to automatically derive a model (or profile) of a person's activities, and then use that model to filter and present appropriate information [13,14]. Example toolsets are OpenIris [15], Nepomuk [16,17], gnowsis [18,19] - a reference framework for parts of the Nepomuk desktop, Mylyn [20] and the related development SmartDesktop [21], and Haystack [22]. A good overview of current developments to enhance the experience offered to PC users is given in [23]. In this section, we describe the key features of a number of these systems.

3.2.1 Mylyn
Mylyn [24] is an Eclipse [25] plug-in that makes interaction and multitasking much easier and reduces information overload. It adds two facilities to Eclipse:

- Integrated task management provides a tasks and bugs tracker and advanced task-editing and task-scheduling facilities.
- An automated context management capability that monitors interaction with Eclipse, automatically identifies information relevant to the task and focuses views and editors to show the relevant information. Based upon Mylyn, TaskTop [26] attempts to bring the same benefits into the desktop, extending Mylyn to email, file and web browsing, improving productivity and alleviating information overload in the Desktop.

3.2.2 SmartDesktop
SmartDesktop uses the notion of projects, folders and associated digital resources (Word files, Excel files); SmartDesktop constantly monitors files and folder associations. Each user can make a project 'active', which explicitly denotes the project he or she is currently working on. If SmartDesktop discovers new resources being accessed, users are prompted to assign them to a project or do nothing. SmartDesktop attempts to present information to the user related to the current project.

3.2.3 Haystack
The Haystack project [22] is developing a platform to allow end users to associate additional descriptions with digital objects. Haystack allows items such as timelines, tasks, and keywords to be related to email threads and documents, allowing associations between arbitrary objects. Thus Haystack allows the end user to access contextual information in any manner that the user requires and the Haystack toolset supports.

3.2.4 Nepomuk
The Nepomuk project [16] is developing a 'social semantic desktop', with associated development APIs to enable further community development upon Nepomuk's knowledge base. The semantic desktop allows a user to annotate and link arbitrary information found on local and remote desktops. Applying peer-to-peer distributed searching techniques, Nepomuk allows members of a community to exchange documents and other tagged pieces of information, as well as information about their individual context and the community's context. Nepomuk provides services and ontologies to model the user environment and the domain of the applications in a standardised and shareable way. This includes a Personal Information Model (PIMO) [27], with the necessary ontologies to represent persons, documents, multimedia, processes etc.

3.3 Informal processes
We discussed previous studies of informal processes and their importance in the knowledge worker's life in section 2.2. However, little appears to have been done to develop software to support such processes or allow machines to 'learn' about them. There is, however, a body of work on the learning of formal business processes. A leading exponent of this is the Process Mining Group at the Technical University of Eindhoven that pioneered the development of the ProM process mining framework [28,29]. More than 200 plug-ins now exist that perform various aspects of process mining via ProM. Typically, process mining is applied to event logs from workflow systems. From this it is possible to discover workflow models and also to achieve process optimisation.

While ACTIVE intends to learn from this work, its situation is less clear-cut. We do not start with a workflow log on which to perform mining. For us it is an open question precisely what events we need to capture in order to perform meaningful mining.

This example describes how ACTIVE technology would work in a particular scenario - that of a patent agent's office. (We have chosen this scenario purely to provide an example that is meaningful to all readers but not based on any of the current case studies. We leave each reader to reinterpret in the context of his or her own work.) The specific process is that of the review of a proposed software patent. The user of the system, John, is a patent agent. The process proceeds as follows:

- John receives a software patent evaluation request by email and sets the current context in ACTIVE Workspace as 'software patent assessment ID 112233'.
- He saves the attachment with the proposed patent and opens the .doc file to read it.
- Meanwhile, he receives more emails. ACTIVE gives them a low priority because, after automatic analysis, they are found to belong to the contexts 'new contract options' and 'software patent assessment ID 112211'. As a result they are displayed in the email client below the emails relating to the current context.
- After reading the proposed patent, John checks a relevant European Patent Office website for policies related to software patents and queries the European Patent Office's database to find any related patents. He also accesses a separate website to read the criteria for evaluation.
- John queries the patent agency's website to determine whether any of his colleagues have filed analogous patents and what he can learn from their experience. To do this he can query on document content and he can also query on the tags created by colleagues. The latter may be more informative. The tags may consist of terms not in the documents themselves but which are particularly meaningful to John and his colleagues.
- John modifies the original word file with his comments and final evaluation.
- Finally, John uploads the file onto the patent agency website and enters some information about the evaluation.

The system takes account of all the resources accessed by John - files, folders, applications, parameters - and creates a context. For this example, it creates a context for 'software patent assessment ID 112233' that can be invoked every time the user is working on this patent. The association of information objects with a particular context will be on the basis of information objects accessed whilst John is in that context and also the content and any metadata associated with the information objects. The system determines that the context consists of a set of files, emails, URLs and applications. Context can also include people - for example, the senders and recipients of context-related emails.

When this process is frequently repeated, the system learns the repeated steps and identifies these as constituting an informal process. So, for example, when searching for related patents and policies related to software patents, John checked relevant web sites and databases. An experienced patent agent will know the appropriate sources of information and steps to take. Because information from one source may be helpful in locating more detailed information in another, the order of accessing the various sources may also be important. Once learned, this process can be easily reproduced in the future and shared with less experienced patent agents.

In addition to developing technology, the ACTIVE project is investigating two important aspects of embedding it in the organisation: the creation and use of cost-benefit models for the deployment of lightweight ontologies and the use of incentives to encourage knowledge sharing. Relatively few have studied the economics of Web2.0 and semantic technologies. Cost models have been developed for the building, reuse and maintenance of formal ontologies [30]. However, little has been done to investigate the costs and the benefits of the Web2.0 approach. ACTIVE deals with the analysis of costs and benefits related to the development, deployment, maintenance and usage of Web2.0 and semantic technologies as they are used to achieve knowledge sharing. One goal is to confirm and demonstrate that using these technologies will result in increased productivity or reduced costs.

We are developing methods to predict and assess the costs of engineering knowledge structures, including modelling, implementation, maintenance and usage, and to show the value for the user and the organisation. Measures will be used to demonstrate tangible and quantifiable benefits within enterprises as an argument to encourage the adoption of our approach. To do so, a model for cost estimation of distributed, collaborative semantic applications and a set of tools to use the model for estimation or planning are being developed.

Furthermore the model will be complemented by benefit estimation measures. A long term goal is to apply known costbenefit rationales to collaborative knowledge creation and elicitation tasks. This will enable the investigation of the impact of cost-benefit related information for decision support - for example, to determine the optimal engineering strategy or degree of formality in collaborative knowledge creation tasks. It is now widely recognised that the corporate culture of an organisation, including its incentives for collaborative work activities, is a critical success factor when it comes to achieving a high rate of knowledge reuse and knowledge sharing. The variable success of collaborative IT applications - given effective and reliable technology - is evidence for the impact of the 'soft' factors. In a corporate context, incentives can be defined formally (for example, monetary and other incentives such as the formal recognition of achievements) but can also be quite informal (for example, enhanced social status and personal recognition within a group of collaborators). When providing content or metadata, for example, individual contributors make decisions weighing the cost of their extra effort against the benefits they and their organisations will obtain. Organisations should provide adequate incentives in their knowledge management (KM) systems to foster the desired form of collaborative behaviour.

We are researching incentives and incentive schemes applied in KM systems and other relevant collaborative applications, focusing on our case study partner organisations and extending to further organisations for which ACTIVE technology may become important. The decisions of individuals concerning their form of collaboration in an environment such as is provided by ACTIVE are made by weighing a number of factors in a dynamic situation against each other, very often under severe time constraints. Experimental research on the effectiveness of the incentive schemes will provide further understanding of the factors which determine the decisions of individuals to contribute to collaborative knowledge systems. The results of our research will be used as the basis of recommendations for incentives to be included in the case study applications. These will be further tested in the field studies.

Within ACTIVE, the applicability of our research results is being validated in three case studies. There are a number of aspects to this validation. At the technical level, we need to understand that the technology works, that the various components work together and that they can be scaled to enterprise size applications. We need also to ensure that the technology is appropriate for users - that is, that it meets their needs and is easy to use. Finally, we need to ensure that the technology is right for the organisation - that is, that it furthers the goals of the organisation, usually in economic terms.

In our case studies, those organisations are relatively large, although much of what we do is applicable to smaller organisations. It is also worth noting that the organisation can be virtual - that is, it might consist of separate legal entities working together for a fixed length of time towards a specific goal. The ACTIVE project has adopted an agile approach which relies on fast design, test and redesign cycles as illustrated in figure 2.

In this process, the design of the component technologies and the application prototypes are steered towards the best fit with the needs and requirements of users and the organisations expected to adopt the technology. During the first year of the project, the information and information system needs of case study users and organisations have been analysed in detail. Throughout the project, components and prototypes are being tested as early as possible as part of an agile development methodology to be certain they correspond to the needs of users and organisations. The case study applications have been validated in short field tests at the end of the first year, and this will be repeated at the end of the second year. A longer period of field tests is then planned for the last six months of the project. The field tests carried out in the three case studies are intended to demonstrate the benefit created by the technology for individual users and user organisations in terms of efficiency of work processes, effectiveness and quality of information. All validation activities are based on a comprehensive set of leading-edge methods which enable efficient and valid assessment of KM technologies. The ACTIVE partners leading the validation tasks are carrying out training activities to make the advanced validation methods available in the IT industry and for technology adopters.

        Figure 2. Agile research and development methodology in ACTIVE

Our particular case studies are within three organisations: BT, Accenture and Cadence Design Systems. Described below, each case study draws on the key technologies being developed and shares common core functionalities. At the same time, each is used to test those functionalities in very different environments. BT needs to support sales and technical staff, helping them respond more rapidly and proactively to customers by making interactions with information systems more efficient. Accenture employs some 180,000 consultants spread over several continents, all of whom need to share its knowledge. And Cadence has a very specialist community of electronics design engineers to support as they navigate and instantiate complex design processes using their experience as a form of tacit knowledge. While each case study has its own characteristics, each provides insight into the generic needs of knowledge workers, and as such is being used to validate technology that is applicable wherever knowledge work is performed.

6.1 Case study: BT
BT Business supplies integrated IT and communications services and solutions to UK-based small- to medium-scale enterprises (SMEs)³ .At a senior management level, there is a requirement for the business to become more responsive to its customers and get better and quicker at turning sales opportunities into contracted business. There is also a strong awareness that, by maximising collective knowledge and skills across BT Business, through, for example, more effective knowledge sharing and reuse, the length of the sales cycle could be reduced significantly⁴. The BT case study therefore focuses on the knowledge transfer and information needs of BT Business's front line sales people – that is, the technical consultants, the solution consultants and the sales specialists working on current sales opportunities. Front-line sales people tend to work on multiple projects and need to be agile in their response to various work demands. In general, there is an increasing demand on the use of their time. As a result, they need to be able to switch quickly between tasks and have all information of relevance to their current task easily at hand. People should not have to search around for such information as they change from one working context to another.

By taking into account the context in which people work, ACTIVE technology will give access to relevant information and knowledge more easily. For example, it will be possible to prioritise the presentation of sales, product and design documentation, and relevant emails. Customer relationship building is another key area of focus. BT's sales specialists and sales consultants are encouraged to build an informed knowledge of their customers – for example, to understand their customers' principal areas of business, their operations, their goals and, where possible, their business plans. Currently, the information that aids this understanding is held in a number of different systems and information sources, and takes time to compile. In order to help the specialist or consultant enhance their customer knowledge, ACTIVE technology will be used to collate customer-specific information according to a person's work-context.

There is considerable potential for people operating in various roles in BT Business to re-use existing solutions, or at least elements of those solutions, in support of new sales opportunities. A better awareness of current solutions to specific business problems and the business domains in which those solutions are being applied should enable common solution patterns to be identified, and enable reuse of previously deployed solutions. This has the potential to reduce the sales cycle time considerably.

Finally, there is a wealth of tacit knowledge across BT that could be used to support new sales opportunities. Currently, this is not being used to best effect. ACTIVE technology will therefore be used to convert tacit and currently unshared knowledge, the so-called 'hidden intelligence' of BT into transferable and actionable knowledge that will support more effective collaboration across the various BT Business units involved in progressing sales opportunities. Knowledge transfer is not just about sharing documents; it is also about identifying people with relevant expertise who can help with a sales opportunity or a technical problem.

---

¹ SMEs are characterised by their diversity, for example an SME could be anything from a start-up or 'micro-business' with 1-10 employees through to a substantial 'medium-sized business' having between 250 and 500 employees.

¹ In the context of this paper, the sales cycle is the period from initial customer engagement to contract sign-off.

6.2 Case study: Accenture
Accenture, a global consulting and outsourcing company, has several application scenarios for knowledge management. Within ACTIVE, we have decided to focus on three where we can achieve more effective knowledge management and thereby improve the sharing and reuse of information that is contained within the enterprise. The first involves a multi-purpose knowledge inquiry. In a large enterprise such as Accenture, a wide range of different functions are performed by a variety of users and it is difficult, if not impossible, to predict and enumerate every task concerned. Even if many of the common tasks can be predicted, it is neither easy nor cost-effective at the moment to build a specialised KM system for each task. Consequently, multi-purpose KM systems are required that can provide information about clients, products, competitors, experts, source code and processes while personalising the delivery of this information dependent on the user concerned and the task in which he or she is engaged.

Although multi-purpose KM systems are important, there is also a need for specialised information access and process support for some of the knowledge-intensive business processes. One such process is proposal writing, which forms the second of the case study scenarios. Accenture currently develops more than 10,000 proposals every year. Currently, this is achieved using a combination of standard processes and a variety of ad-hoc mechanisms that varies from user to user. The ad-hoc mechanisms involved include:

- calling contacts that may have information on similar projects;
- using search to locate similar proposals; and
- using instant messages and/or emails to gather additional information from colleagues and friends.
ACTIVE technology will be used to:

- support collection of the necessary information to write proposals;
- give users access to 'fine-grained content' instead of a set of documents through which they must read to find relevant content;
- provide workflow and process support for collaborative proposal writing; and
- semi-automatically review proposals for compliance, consistency and the integration of key themes that have been deemed necessary for the proposal.

The third scenario is based around a portal-like integrated system for information access. The Accenture marketing practice frequently designs marketing information for specific clients. For instance, a variety of marketing material may accompany a proposal that is submitted to a client. Such marketing material will commonly include information about previous projects that Accenture has done for the same client, similar credential information and general information about Accenture and its successes. There are two main areas for improvement to this process.

The first is to locate information. Often, information about previous projects cannot simply be obtained from the central Accenture data repository. This is because many documents created within the context of a client project are clientproprietary and may not be shared with all Accenture personnel. The necessity to restrict certain information means that information is not available to those who could legitimately use it. The second challenge is the duplication of work. Within marketing, many client-specific artefacts are created that could be reused in other projects. This, however, is not always done, simply because one person does not know that somebody else has already created a very similar artefact.

6.3 Case study: Cadence
Cadence Design Systems GmbH is an engineering design services provider that operates in the domain of microelectronics and integrated circuits. This case study deals with eliciting tacit design process knowledge from the everyday practice of design engineers in this engineering design domain. Though design technology in this field is well defined, the processes still remain very stochastic. Some important reasons for process dynamics are:

- that design goals may change in the course of a process;
- that the designers involved may have different skills, capabilities and tool preferences; and
- that even the order of some phases (activities) cannot be pre-set deterministically for every possible type of an artefact under design.

A design process therefore develops dynamically. At every step, a decision needs to be made as to which of the admissible continuation design paths is more productive. The challenge is that the knowledge required to make such decisions is tacit, subjective and cannot be fully elicited or made explicit by interviewing designers. In particular, designers often employ their intuition and experience when reasoning about facilitation dependencies among design activities situated in design environments.

Our approach in ACTIVE is to record the performance of engineering design processes in the everyday practice of experienced designers, taking account of all relevant legal, ethical and trade union considerations in the process. Several concurrently and autonomously executed design processes are tagged by the ACTIVE software components incorporated in the Cadence Workbench software. The components produce process information in the form of 'ticks'⁵. These ticks are stored in the knowledge base.

Many design engineers, distributed over several development teams and several engineering design processes, will use the 'ACTIVated' Cadence Workbench in their work. The workbench is intended as a collaborative engineering design process tool permitting process tagging. Process mining components use the 'ticks' knowledge base to understand the sequences of atomic activities and compound tasks executed in the processes. Process execution paths are thereby recorded in the knowledge base.

At each point of decision making, a designer has to choose the most productive continuation path for his current process using activity or task patterns prescribed by the technology or in-house policy. Several admissible alternatives may appear to be possible. For those, advice is solicited from the ACTIVE software. The system will suggest the most highly-ranked continuation path appropriate for the context of the current process using performance comparisons of other processes in similar contexts, already recorded in the past.

---

⁵ A 'tick' is essentially a process event. The word 'tick' is preferred to 'event' to avoid confusion with other uses of 'event'.

We began this paper by observing that, over the last few decades, knowledge workers have been given many significant new ways to access and process information, but that these new capabilities have brought new problems. Even within an organisation, knowledge workers rarely share the information that is available as fully as they might. They can be overwhelmed with information that is of varying degrees of relevance. And they can struggle to interact with complex systems, constantly reinventing their own informal processes as they do so.

We believe these problems can be most effectively addressed through a synergy of Web2.0 and formal semantic technology, through the use of contextual understanding to guide information delivery, and through support for informal processes.

We have explained how, to confirm these assumptions, ACTIVE is not only undertaking a programme of research and technology development, but also comprehensively validating its technical results in three case studies. Our end goal is to very significantly enhance the experience offered to knowledge workers in all sectors of the world economy.

Rather than developing new applications for users, ACTIVE is developing a framework to support the use of existing applications. That framework will help users to easily share information, will deliver information according to users' current task context, and will support the reuse and sharing of informal processes. ACTIVE is being validated in three case studies, the results of which are being continuously fed back into the development process. The success of ACTIVE will be judged, from both the user's and organisation's standpoint, within those three case studies. The project runs for three years until February 2011 and we expect to report on both the progress of the project and the results of the validation exercises regularly throughout that time period.

The work described in this paper has been funded as part of the IST-2007-215040 EU project, ACTIVE.

1. ACTIVE web site, http://www.active-project.eu Top
2. Drucker PF, 'Management's new role', Harvard Business Review, 1969, pp.49-54 Top  
3. Davenport TH, 'Thinking for a living', Harvard Business School Press, 2005 Top  
4. OASIS, 'WS-BPEL - the web services business process execution language', http://docs.oasis-open.org/wsbpel/2.0/wsbpel-v2.0.pdf (accessed December 2008) Top
5. The Object Management Group, 'Business Process Modeling Notation', http://www.bpmn.org (accessed December 2008) Top
6. Davies J, Studer R, Sure Y and Warren P, 'Next generation knowledge management', BT Technology Journal, vol.23, no.3, 2005, pp.175- 190 Top
7. Flickr web site, http://www.flickr.com Top
8. Völkel M, Krötzsch M, Vrandecic D, Haller H and Studer R, 'Semantic Wikipedia', 15th International World Wide Web Conference, WWW2006, May 2006 Top
9. Sixth International and Interdisciplinary Conference on Modelling and Using Context, Context 07, http://context-07.ruc.dk Top
10. Hill C, Yates R, Jones C and Kogan SL, 'Beyond predictable workflows: Enhancing productivity in an artful business process', IBM Systems Journal, vol.45, no.4, 2006, pp.663-682 Top
11. Hayman S, 'Folksonomies and tagging', Ark Group Conference: Developing and Improving Classification Schemes, Sydney Australia, June 2007, http://www.educationau.edu.au/jahia/webdav/site/ myjahiasite/shared/papers/arkhayman.pdf Top
12. Kings NJ, Gale C and Davies J, 'Knowledge sharing on the semantic web'. In Franconi E, Kifer M and May W (eds.), 'European semantic web conference', Springer-Verlag, 2007 Top
13. Crossley M, Kings NJ and Scott JR, 'Profiles - analysis and behaviour'. In Ralph D and Searby S (eds.), 'Location and personalisation: delivering online and mobility services', IEEE, London, 2004 Top
14. Jones W, 'Keeping found things found', Elsevier, London, 2008 Top
15. Cheyer A, Park J and Giuli R, 'IRIS: Integrate. Relate. Infer. Share', Proceedings of the Workshop on the Semantic Desktop: Next Generation Personal Information Management and Collaboration Infrastructure, Galway, Ireland, 2005 Top
16. 'NEPOMUK - the social semantic desktop', http://nepomuk. semanticdesktop.org Top
17. Groza T, Handschuh S, Moeller K, Grimnes G, Sauermann L, Minack E, Mesnage C, Jazayeri M, Reif G and Gudjonsdottir R, 'The NEPOMUK Project - on the way to the social semantic desktop', Proceedings of ISemantics 2007, pp.201-211 Top
18. DFKI Knowledge Management Lab, 'gnowsis', http://www.gnowsis.org/ Top
19. Sauermann L, Grimnes G, Kiesel M, Fluit C, Maus H, Heim D, Nadeem D, Horak B and Dengel A, 'Semantic Desktop 2.0: The gnowsis experience', Proceedings of the ISWC Conference, 2006 Top
20. Kersten M and Murphy GC, 'Foundations of software engineering'. In Proceedings of the 14th ACM SIGSOFT International Symposium on Foundations of Software Engineering, 2006 Top
21. Decho Corp, 'Smart Desktop', http://www.smartdesktop.com Top
22. Karger DR, 'Haystack: per-user information environments based on semistructured data'. In Kaptelinin V and Czerwinski M (eds.), 'Beyond the desktop metaphor: designing integrated digital work environments', MIT Press Books, 2007 Top
23. Kaptelinin V and Czerwinski M, 'The desktop metaphor and new uses of technology'. In Kaptelinin V and Czerwinski M (eds.), 'Beyond the desktop metaphor', The MIT Press, 2007 Top
24. The Eclipse Foundation, 'Mylin', http://www.eclipse.org/mylyn Top
25. The Eclipse Foundation, http://www.eclipse.org Top
26. Tasktop Technologies Inc, 'Tasktop', http://tasktop.com/ Top
27. Sauermann L, van Elst L and Dengel A, 'PIMO - a framework for representing personal information models'. Pp.270-277 in Pellegrini T and Schaffert S (eds.), 'Proceedings of I-MEDIA '07 and I-SEMANTICS '07 as part of TRIPLE-I 2007', Graz, Austria, September 5-7, 2007 Top
28. The Process Mining Group, Eindhoven Technical University, 'The ProM framework', http://prom.win.tue.nl/tools/prom/ Top
29. van Dongen B, de Medeiros A, Verbeek H, Weijters A and van der Aalst W, 'The ProM framework: a new era in process mining tool support', Lecture Notes in Computer Science, vol.3536, pp.444-454, Springer Verlag, June 2005, Top
30. Simperl E, Tempich C and Sure S, 'ONTOCOM: a cost estimation model for ontology engineering', Proceedings of the International Semantic Web Conference ISWC2006, Springer Verlag, 2006 Top

Paul Warren works in BT's centre for information and security systems research where his interests centre on knowledge management, semantic technologies and the service oriented infrastructure. He is currently Project Director of ACTIVE (http://www.active-project.eu), a European project in the area of collaborative knowledge management. At an earlier stage in his career, Paul worked on technology strategy and technology foresight, investigating areas as diverse as e-business and novel forms of computing. His previous experience also included a secondment to the Confederation of British Industry to study government support for technology in industry. Paul has published widely on knowledge management, semantic technologies and technology foresight, and has collaborated in editing two books on topics related to ICT. He holds a degree in theoretical physics from Cambridge University and an MSc in electronics from Southampton University.

Nick Kings joined BT in 1984 after gaining a BSc in Computer Science and Electronic Engineering from Birmingham University. Since then, he has worked on the development of advanced automatic software development systems, the visualisation of complex software systems and ways of improving team-based software development. Now a member of the nextgeneration web research group in BT's centre for information and security systems research, he has developed tools to support knowledge sharing in online communities and is currently researching the applications of semantic web technology to knowledge management, information retrieval and communities. Nick gained an MSc in telecommunications engineering from University College London in 1997 and is currently working towards a PhD based on his work in knowledge sharing, formal ontologies and informal 'folksonomies'.

Ian Thurlow is a researcher in BT's centre for information and security systems research. Since joining BT (then Post Office Telecommunications) in 1978 as an apprentice, he has been awarded a BEng in electronics engineering by the University of East Anglia and an MSc in telecommunications engineering by University College London. Now a member of the Institution of Engineering and Technology and a Chartered Engineer, Ian has extensive experience in managing research and software development projects and was involved in the formation of Exago, a knowledge management company that span-out of BT in 2000. He managed BT's case study contribution to the EU Sixth Framework project, Semantically Enabled Knowledge Technologies (SEKT), and is leading BT's case study work on the ACTIVE project.

John Davies leads BT's next generation web research group and is chief scientist, semantic technology in BT's centre for information systems and security research. His interests include the application of semantic and Web 2.0 technologies to business intelligence, information integration, service-oriented environments and knowledge management. John has written and edited many papers and books on semantic technology and its applications, web-based information management and knowledge management. Currently co-organiser of the European Semantic Web Conference series and chairman of the European Semantic Technology Conference, he has served on the program committees of many conferences in related areas. John is a Fellow of the British Computer Society and a Chartered Engineer.

A researcher at STI Innsbruck, Tobias Bürger is currently studying for a PhD at the University of Innsbruck. Before joining STI, he was a project manager and researcher at Salzburg Research, working in the area of digital media. His primary domains of research are ontology economics, semantic web technologies and multimedia semantics, fields in which he has published around 35 scientific papers. An active contributor to EU collaborative projects, he also contributes to the World Wide Web Consortium's work on cost and benefit estimation for ontology-based applications.

Elena Simperl is the vice-director of the Semantic Technology Institute (STI) at the University of Innsbruck. She holds a PhD in computer science from the Free University of Berlin and a diploma in computer science from the Technical University of Munich. Her research centres on knowledge engineering. In particular, she is interested in user- and business-oriented aspects of ontology building and management, and has approached these topics in over fifteen national and European research projects. She has also organised scientific workshops on these topics alongside the world's leading conferences on semantic technologies. Since 2006, she has been the instigator of events such as the PhD Symposium, held in conjunction with the European Semantic Web Conference.

Rayid Ghani leads the analytics research group at Accenture Technology Labs. His research interests include the application of machine learning, data, text and web mining to current and emerging business problems. His recent work has centered on knowledge management and data anonymisation. Rayid has organised workshops at the International Conference on Machine Learning, the Knowledge Discovery and Data Mining Conference and other conferences, bringing researchers from academia and industry together to share insights and ideas in the fields of machine learning and data mining, and has published papers in leading journals and conference proceedings.

Carlos Ruiz is a researcher at Intelligent Software Components (iSOCO) in Madrid. He holds a degree in computer sciences engineering from the Technical University of Madrid and is currently finishing a PhD in the field of semi-supervised clustering and data mining. In 2001, he joined the CETTICO (Department of Computer Technology Transfer) research group at the university, where he was enrolled in some national projects regarding human-computer interaction and artificial intelligence as software engineer and researcher. In 2007, Carlos was a lecturer in the Computer Architecture and Automatics Department at the European University of Madrid. During his career, he has published articles in books, journals and conference proceedings on subjects including artificial intelligence, data mining, computers for people with special needs and human-computer interaction.

Marcel Tilly received his masters degree in physics from the Technical University of Dortmund. With more than 10 years of developing and consulting on software engineering, he has made significant contributions to the technical architectures of several large distributed software systems. Marcel is co-author of a book about web development and has appeared as an invited speaker at several conferences. In 2006, he joined the European Microsoft Innovation Center in Aachen as a program manager, where his work focuses primarily on event processing, service composition and user behaviour analysis. His current work in ACTIVE is related to context-aware knowledge processes.

Jose Manuel G�mez-P�rez is a research manager at Intelligent Software Components (iSOCO) in Madrid. Working both at iSOCO and the Technical University of Madrid, his research has focused on the semantic web, knowledge acquisition, representation and reasoning (especially with process knowledge) and the analysis and interpretation of process executions. Jose Manuel has contributed to numerous international research programmes, leading iSOCO's contributions to projects like NeOn (http://www.neon-project.org), ACTIVE (http://www.active-project.eu), OntoGrid (http://www.ontogrid.eu) and Halo (http:// www.projecthalo.com), a project funded by Vulcan Inc. Keen to see the technologies he has worked on transferred into industrial sectors including telecom, automotive, finance and public administration, he is a regular contributor to conferences and journals, has contributed material to several books and has helped organise a number of scientific and industrial events.

Tom Bösser is the founder and director of keapro GmbH, a business that researches and consults on user analysis for innovative technology. After studying psychology, physiology, philosophy and applied mathematics in Konstanz, London and Düsseldorf, he became a researcher and professor in Düsseldorf, Darmstadt and Münster, a visiting scientist in Oxford, Paris and Berkeley, and a research manager in the electronics and IT industry.

Vadim Ermolayev holds a PhD in computer science from the University of Zaporozhye, Ukraine. He is a research consultant at Cadence Design Systems GmbH and an associate professor in the Department of IT at Zaporozhye National University. The founder and leader of the university's intelligent systems research group, his research interests centre on knowledge engineering, agentbased computing, process and performance modelling, the semantic web and services.

Ali Imtiaz is a senior researcher in the information management department of the Research Institute for Operations Management, Forschungsinstitut für Rationalisierung (FIR), in Germany. Building on a diverse academic background spanning business finance and information and communication technology, his practical experience covers both business administration and information technology adoption. At FIR, he has worked on a number of European collaborative projects as a project manager or researcher. His research interests centre on business models, information technology and systems.

« Previous

|

|