Raven was always going to be an important product, not only for Lotus, but also for the knowledge management market in general. The release version of Knowledge Discovery System shows Lotus has not backed away from its original vision and ambition - even if that has meant taking some hard knocks over delivery dates.
There are two key components to the system: K-Station and Discovery Server. K-Station is the portal component, and was released in late 2000. Discovery Server (released at the end of April) provides taxonomy generation and expertise location services.
Together, as the Knowledge Discovery System, they provide a comprehensive knowledge management tool set that can be tailored to a wide range of vertical and horizontal applications. This brings into the mainstream market advanced knowledge management features that will both benefit and challenge existing managerial and knowledge management practices.
Taxonomy creation and management will be a new process for many organisations that have never seriously addressed the issue of information classification on a corporate scale. The proliferation of intranets is forcing many companies to face up to this problem for the first time - Discovery Server provides an automated solution. But while the product automates much of the process, to get the maximum benefit organisations will need administrators, information managers or "cybrarians" who understand the challenges of corporate information management.
Today, few organisations have the management processes or culture in place to match the capabilities of such technology - and their willingness to develop those skills will be crucial to the success of such products.
These challenges are not limited to users of Lotus's products - they are relevant to all companies trying to provide sophisticated knowledge management support. If we are to take knowledge management to the next level - as a serious discipline aimed at the effective use of intellectual capital within organisations - then products such as Knowledge Discovery System (and yes, Microsoft's SharePoint too) will need to become accepted components of the corporate infrastructure, and not merely passing fads.
Lotus first started talking about its advanced knowledge management tools in 1997. Document classification tools were originally announced as part of the R5 release of Domino - but they were then held back and rolled into a distinct product, codenamed Raven.
Lotus has emphasised throughout this long wait that it would not release the product until it was ready. It now has to live up to that claim (and to heightened expectations about functionality) when customers start to go live with projects.
Lotus says the delays were largely a result of work on usability improvements, performance enhancements and, in particular, on the need to assure that the taxonomy generation tools work effectively. During Raven's long haul to actual availability, Lotus's vision for the product has remained remarkably constant. Back in 1999, the first mock-ups of the product focused on the theme of people, places and things, and this has been the mantra for both the development and marketing teams ever since.
Lotus K-Station is a portal environment that allows users to define and work within a variety of shared "places". The idea is that places support the multiple roles that each user takes on each day and as their career develops. K -Station is a key element in what Lotus calls the move to the "occupational-centric" desktop. It allows users to define their own (personal or communal) places from sets of shared "portlets" that provide access to commonly used applications and data sources.
A user can navigate (via tabs or bookmarks) from a home site, My Place, to places for the various communities to which they belong - for example, the senior management team, product development, specific projects or recreational communities, such as a running club. Some places are defined by the user - including their own home site and any others for which they are responsible as administrator or manager. For other places - for example, departmental sites - they may take some, but not sole responsibility for definition and development.
Lotus has tried to make the creation of new places a simple task, with as much automation as possible. There is close integration with its QuickPlace product, so existing QuickPlace spaces (for example, for a project team) can be incorporated in part or as a whole within a K-Station workspace.
Lotus Discovery Server will appeal to large organisations attempting to bring order to proliferating document stores, intranet sites and dispersed teams of experts. The Discovery Server combines categorisation, taxonomy generation and expertise location services in one product. Unlike some other categorisation tools, Discovery Server is very much a corporate product. It is designed to help organisations provide a common taxonomy of information that can bring order to a user's search for documents and people.
At the heart of Discovery Server is the Discovery Engine, which analyses documents and other information in order to generate a classification model or taxonomy. What distinguishes it from most search tools is that it can categorise the relevance of documents, people, and of existing K-Station places.
The starting point for using Discovery Server is to create a K-Map or corporate taxonomy. An existing taxonomy can be used as a starting point - or Discovery Server will generate a taxonomy from a sample set of documents. This first cut can then be refined using the K-Map Editor - so an administrator or knowledge manager can assign appropriate labels to categories or move documents to other categories. Once they are happy with the model, it can be used to categorise documents and data sources automatically.
The expertise location services in Discovery Server are particularly important. The Discovery Engine creates a profile of a user by drawing on various sources including:
- existing directories (either Domino or LDAP)
the quality and quantity of relevant contributions made by the user to the corporate repository
manually maintained information (such as interests, education and employment history).
Using this information, Discovery Server identifies a user's areas of expertise or "affinities" in so far as they relate to the categories in the K-Map. Lotus has addressed the privacy issue, which is always raised in relation to any user profiling by giving users control of their own profile. A user can remove affinities or add new ones (but their score will be set low initially); they are notified if a new affinity is added to their profile; they can also decide whether to provide additional information in terms of education history, interests and so on.
The user's view of Discovery Server is driven by a normal text search entry - but it is the results that are different from normal search engines. Results are grouped into three types of resource - documents, people and place. Document result sets are shown in a familiar search results screen - including a document summarisation option. The results of the people search identify "experts" that have a close affinity with the topics being searched for. The places section shows existing K-Station places that are closely related to the search term.
The expertise search is particularly impressive when combined with Lotus collaboration tools. So, for example, having identified an expert on a chosen topic, you can then use the Sametime instant messaging functions to see if the person is online and, if so, start an immediate dialogue with them.
Some aspects of Discovery Server are disappointing, however. There is, for example, little exploitation - from a management perspective - of the metrics being collected on expertise and document usage. In addition, the fact that Discovery Server has been designed as a corporate top-down product means there is little in the way of user customisability. For example, it would be useful to be able to define your own personal views on the taxonomy, or generate your own versions based on a sub-set of documents. But these are minor faults in the first release of an impressive product.
Not surprisingly, Lotus is looking to its business partners to develop horizontal and vertical applications based on the product. But there is also significant potential for development of the core product itself. Most tantalising is the potential for surfacing more of the data being collected by Discovery Server as part of the classification process. For example, the metrics collected on people and documents could be used as the basis for a comprehensive corporate skills audit.
The team responsible for developing the IBM/Lotus e-learning solutions is also excited about the possibilities of integrating the formal e-learning options offered by LearningSpace, with the informal learning that could be supported through K-Station and Discovery Server.
Another important - if more contentious - area for future development is to broaden the product's appeal beyond the existing Lotus customer base, which is obviously the initial target audience for the products. The Knowledge Discovery System is also an important weapon in Lotus's battle with Microsoft for large enterprise accounts. (It is notable that merger and acquisition deals that force organisations to decide between existing Domino and Exchange infrastructures are becoming one of the most hotly contested parts of this market.)
Lotus has always claimed that Raven was designed to appeal beyond customers who have already invested in a Domino infrastructure. In theory, there is no reason why the Knowledge Discovery System has to run in a Domino environment. Although at least one Domino server licence is required at present, the product can already integrate with an Exchange-based messaging infrastructure. However, to seriously build a non-Domino customer base for the product, Lotus will have to address a number of stumbling blocks.
It will have to reduce (or hide more effectively) the Domino bias in the product. While optimisation for Domino/Notes is understandable and attractive to traditional customers, it will be a negative for other customers. The way to address this is to beef up support for non-Domino infrastructures.
The fact that it uses a DB2 database to store catalogue and metric data (albeit as an under-the-cover component) will also be a negative factor for many IT shops with a religious attitude to their chosen database supplier. However, given that non-Lotus IBM customers are inevitably the next target audience, this may not be a concern for the medium term.
More than any technical issues, there are the cultural barriers to selling outside the existing Lotus market. Lotus will look to the IBM salesforce to pitch Knowledge Discovery System into accounts that are closed to Domino as a messaging solution. We believe this will be a tall order for the company, but if it succeeds, Raven really will have been the start of a revolution at Lotus.
Dr Eric Woods is research director, knowledge management, at Ovum.
Ovum, the analyst and consulting company, is a global leader in the rapidly changing world of converging technologies and markets. Resolutely independent, Ovum offers clients objective advice and challenging insights.
This was first published in May 2001