Intranets become more and more like small Internets with hundreds of thousands of documents inside. This is especially true for multinational enterprises. The Internet itself, with its extremely fast growing number of documents (a few billion) is now the biggest available source of information on earth.

Nevertheless, information held by employees of an organisation is even more worth full. The right information at the right time and the right place is one of the biggest assets of companies. Therefore it has to be determined, which member of an enterprise holds which knowledge. With the complicated hierarchy structures of multinationals nowadays, this information does almost not exist on a big scale and would be extremely valuable.

Consequently new approaches and technologies are needed on how to retrieve documents and on how to determine who holds this knowledge, which are beyond current existing technologies. The here presented breaking edge technology is probably the first in finding a solution to these upcoming problems.

OKDOO presents the first automatic knowledge management system and personalised search engine. To illustrate this, one can imagine a child that is searching for information to solve his homework and a physician. If both would enter the same keyword "physics" in current search engines, both would get the same results. OKDOO instead would present the one introduction to physics, while the other would get the latest scientific results concerning his field of application in physics.

This is done, by automatically creating very detailed profiles of users, while they are using this new technology. While developing this system, it was made sure, that privacy of users is respected.

Following you will find a motivation and a brief introduction in the technology, followed by a description of products that are directly derived from the core system.

To enable detailed and valuable results for a query, done by a specific user with specific skills, knowledge, interests, culture, etc., the system follows a new approach. It consists of a self-learning technology, which makes extensive analysis of the users behaviour, to learn for example in which topics he is interested in or what skills he has. Users with similar properties in similar topics will be grouped together, which means, that users accessing a specific subject will get much more precise results, as long as they belong to a group where other already rated documents as being valuable. This complete process is absolutely transparent to the user. In fact, the only thing he has to do is to use the system like a usual search engine. The technology will automatically handle the grouping of users and the rating of documents.

The first step: To be able to show to the user the most appropriate information, he needs to be identified (as far as possible). This could either happen by login and passwords, or by cookies, IP-Addresses and an intelligent pattern matching.

The second step: After the user has formulated his request, by entering a query, the system will check his profile and crosscheck it with similar profiles to then extract a number of documents, that will match the users demand. These documents will then be ranked and presented to the user.

The third step: Successively a personalised directory of documents will be created ordered in a hierarchical and/or chronological way. This would mean (if the necessary data is available), that users, groups and organisations could get their up-to-date and own directory without deploying any effort.

On top of this, users, groups and even organisations or groups of organisations can have an overview on the mainstreams and topics of interests in an easy accessible overview.

Each computer having access to the Internet and a standard Internet Browser (like Netscape Communicator or the Microsoft Internet Explorer) installed can use the System, without the need to install any other software.

The technology is accessible, via a web address and used like a normal search engine, where queries are inserted in a text field and, by pressing the submit button, a result set of documents will be presented. The presented results will be personalised, adapted to the profile of the user. The more a user accesses the services, the better the results he will get.

It is planed to provide a personalised directory that will probably be presented in a YAHOO-Style Web Directory. This hierarchical and topic structured directory, will give easy access to the user’s recently visited documents and/or important documents. Additionally the user will be able to chose a directory of user groups. Ex.: All documents that are and were important for the Atlas Collaboration.

Users can register on the System, but are not obliged to. Registering will enable the functionality to have the same profile visible on other computers (like a home computer) transparently.

The services of the new technology are provided on the basis of a client/server structure. The client is represented by the computer, which a user uses. The client accesses the server, which runs the technology, via a standard TCP/IP network connection using a standard Internet-Browser.

The Server itself consists of a 3 tier layer, were each layer can consist of computer clusters. The first "presentation layer" handles request of users and dispatch these requests to the appropriate computer(s) and prepares the reply to the client. The second "application layer" does all the necessary calculations, to treat profiles, groups, retrieving and ranking of documents. The third "database layer" runs a database server that enables to retrieve and store large amount of data in an efficient way. Currently all Software needed to run the server is based on a very cost effective "open source".

The core of the technology is base on a new type of artificial intelligence, using implicitly the intelligence of the users. A more detailed description will be available very soon in my doctoral thesis called "Efficient Graph Patter Matching through the usage of implicit semantic Structures".

The storage and processing of user profiles is done using graphs. These will have to be matched against each other, to determine the proximity of users. Before the invention of my technology the main drawback of graph matching stems from the fact that it belongs to the class of NP-complete problem, which means that with the number of existing profiles, the complexity of processing these profiles increase with an exponential factor. Therefore, the method may grow out of computational tractability in the worst case. Using implicit semantic structures I invented a solution to this problem, which brings the complexity down to a computable level achieving a very high performance.

Derived from the core system one can imagine lots of different applications possible on top of it. Using the integrated XML Interface the technology allows connecting to almost every other modern application. In the beginning the product framework shown in the figure below will be provided:

With the detailed profiles of users, OKDOO can automatically provide forums, in which members of an organisation can share, their knowledge and ask questions. These forums will be built on specific subjects, to fulfil the need of the organisation. The administrative work to manage these tools is almost zero, due to the fact, that the system has the ability to generate such forums on simple demand. Of cause an administrator would get the possibility to define which groups of the company will have access to which part of the knowledge structure of the organisation itself.

Also OKDOO provides a facility in which collaborations can be defined, to share knowledge with. Taking CERN as an example, one could imagine that a forum can be shared with Institutes all over the world that are working on the same project. The Forum itself can have a hierarchy structure, to show different problems that are part of the realisation of the project. Users will have the possibility to select if they want to get Mails on new entries in a board of a forum or even to select, from which persons they would accept mails.

Current search engines do not make any difference between users. Let’s look again to the example already mentioned above. If for instance a high skilled physician makes a search with the keyword "physics" and a child that is searching for information to do his homework would do the same query, both would get the same results on conventional search engines. OKDOO with its new cutting edge technology will provide different results. The child would in this example get introductions to physics, while the physician would get the latest results of scientific research.

With the massive grow of documents in Intranet of organisations and the Internet with its billions of documents available such new high-end technologies will be a necessary tool. It is planed to give organisations licences to use this technology for their needs, which can be chosen, out of different levels. One big advantage is the very simple introduction of this self-learning technology.

OKDOO also introduces the first real personalised directory, which will be built automatically on behalf of the user’s interest. Users will be able to see a directory of documents that can be structured in different ways (similar to the one known from Yahoo). The directory views can be chosen by interest, by groups the user belong to and by the organisation. With this, users will have an easy access to their favourite information as well as to important documents of their division/groups and those of the organisation, which will held them up to date any time and any where.

The knowledge and skills of the members are one of the biggest assets of an enterprise or organisation. To efficiently access these skills, the organisation has to know which members or group of members has the desired skills to find solutions for existing problems.

Nowadays such knowledge maps (comparable to a roadmap, but instead of locations, members are positioned) almost don’t exist, often because of rigid and complex structure of organisations. One can imagine, that in multinational enterprise, where different division in different countries can encounter similar problems quite often. Daily work has proven that the necessary information to solve a specific problem is not found within a short amount of time. This will result in lots of time spent to reinvent the wheel (even, if a solution already exists inside the organisation that is not found because the necessary tools are not available). OKDOO provides probably the first tool, filling this gap.

While using OKDOO, the system implicitly learns from every user and step-by-step build a very detailed profile of the interest of users. In the environment of an organisation, this information can then easily be used, to build communities for instance for new projects.

The Internet advertising market experienced a big decrease in ad-revenues the last 2 years. This was due to the insufficient reach of the target consumers. Advertisers nowadays want to reach their possible customers with a very high precision, to be willing to spend money on advertising. OKDOO’s profile based system can enable such targeted groups for the free-for-all internet search engine. This can bring some additional revenues. Moreover, with the analysis of the requests done by users, it can provide a transparent market analysis for various fields, and this in an automated way.

As a further product in OKDOO’s framework the system can act in two ways. As an external provider of content for almost any type of portal system. On the other hand, it can also be used to group Intranets of large organisations together, to give users a transparent access to all data available over different Intranets of one or more Organisations. This usually happens for multinationals, which have their own Intranet in the different dependencies in different countries.