Reference Projects

Semantic search for the technology portal of civil engineering company Bilfinger Berger

Text technology for the multimedia Brockhaus Encyclopedia

Recommender systems for  idea management

Joint development project for an intranet applications.
Semantic search for the in-house technology portal of civil engineering company Bilfinger Berger AG.

The international corporation Bilfinger Berger makes a portal available on its intranet
to its employees with specialist information about structural engineering and related areas. This knowledge is stored in distributed databases.

Fraunhofer IPSI is developing an innovative system in a current project with Bilfinger Berger which integrates distributed, multilingual data sources and makes structured access to them possible by means of a semantic search. The knowledge network planned for Bilfinger Berger has the task of indexing the content of the databases of the technology portal and thus making an integrated search of them possible. It will consist of various logical partial networks: partial networks for information types, individuals, keywords, and topics which support semantic search. Computational, linguistic and statistical techniques will be employed by Fraunhofer IPSI for the automated creation of the knowledge network. Periodic updating of content also takes place automatically.

Engineers on site or employees in the office can access all the data with a single search query and have the results presented to them in a clearly laid out form, structured according to content. For example, even though the information comes from different sources, they can find out with one query, for example, what special expertise an employee has, which projects were carried out, which innovations emerged or were used, which norms and guidelines were applied or which patents exist.

Text technology for the multimedia Brockhaus Encylopedia

ORION is conducting computational linguistic analysis of encyclopedia and dictionary entries in the context of the project  "Dictionary and encyclopedia based knowledge acquisition" for the publishers Bibliographisches Institut & F. A. Brockhaus AG. This analysis extracts relvant semantic information from the texts. The aim of the project is to allow normal users to support their searches in full text databases with natural language and dialogue-based search technologies, navigation in a 3D visualisation of knowledge spaces., and computational analysis of text. The project results will be realised in a prototype application in the "Brockhaus multimedial" multimedia encyclopedia.

Recommender systems for idea management

Fraunhofer and HLP introduced recommender systems for idea management at CeBIT 2004

The tried and trusted "suggestions box" in companies has long since grown into idea management such that the considerable potential of employee suggestions  for innovation and optimising business processes can be fully exploited. Idea management systems support every stage, from brainstorming right through to awarding prizes for particularly successful suggestions. Idea management systems can easily have to handle a hundred thousand suggestions in large companies - a figure which makes it hard to separate the wheat from the chaff and to identify suggestions for improvement which exactly address a specific problem.

In 2003, employee ideas led to substantial cost reductions of 169 million Euro for Siemens, according to the German Institute for Business Administration (dib) in Frankfurt. Across Germany the institute was able to identify around 1.2 thousand million Euro savings as a result of idea management.

Up to now, however, suggestions for improvement have largely only been taken on board in the department in which they were made, whilst other parts of the company have not been able to benefit. How can the whole enterprise, which may operate internationally, have full access to the whole pool of ideas?

The solution is a recommender system which the Orion division of the Fraunhofer Integrated Publication and Information Systems Institute developed in collabortion with HLP Informationsmanagement GmbH, a provider of  IT services based in Eschborn, Germany. The system was presented at CeBIT 2004 in Hannover on the Fraunhofer booth.

The recommender system analyses similarities in suggestions for improvements and is thus able to spot relevant ideas and to put them forward as suggestions, even when these ideas originally appeared in a completely different problem scenario. The system also independently determines the potential for improvement of indvidual departments and proactively puts together a list of relevant suggestions for improvement. A search in the ideas database provides a larger number of meaningful results than would a conventional full text search. Furthermore, users of the idea management system are encouraged to adopt ideas from other departments and other usage contexts. The recommender system for idea management promotes the spread of suggestions for innovation and optimisation not just across the enterprise but also with relevance to individual departments. The results are clear to see: companies can make substantial savings with the help of this system. The technology also serves to motivate employees who put forward ideas, since they can observe with pride how their idea is put into practice across company locations.

The recommender system runs within a Java application server and can be seamlessly integrated into the enterprise's SAP strategy in combination with HLP's SAP-certified idea management system. Companies which are not SAP users can, however, still make use of HLP idea management and the new recommendation system.