30.000 advisory calls per year! How to analyze, rate and improve them interactively? For getting these goals a novel technique is introduced, which is based on a structure similar to a music score. Via a “duet”, guided by the advisor called, the main contents of the dialogue are extracted and stored automatically. At this, trained speech recognition is applied for the (known) advisor’s part for monitoring and controlling the call’s course, and keyword spotting is used for the (unknown) “customer” to pick out instances from sets of expected values. The results thereof fill an abstract score annotated to the call conducted. Then, because of the strong structure of such scores, the archive of all these dialogues is an ideal base for retrieval, exploration, comparison and clustering of calls, using e.g. graphical tools like knowCube®.
There are many technologies calling themselves knowledge management systems in the market but most of them deal only with information management. The main difference between them is that knowledge management systems are oriented into people and participated in tacit knowledge capturing. The paper describes set of information technologies which could take part in all process of knowledge management and especially in collecting tacit knowledge.
Classical approaches to document management do not cope with the demands knowledge documents make. Knowledge documents or so-called “living documents” have a far more complex lifecycle than general documents. They are usually used, edited und utilized by several people and many different versions, revisions and variants exist. Today the multitude of information that these processes generate are not captured or used to provide a better management or retrieval for this kind of documents. Our approach shows that the capturing of this lifecycle information can help in the retrieval as well as usage and management of those documents.