Figure 1-1. Outline of the BRIEFS 4 System.
The BRIEFS System processes documents of text. Its intention is threefold: to support, based on common base of linguistic and statistical analysis,
Extracting computable data for knowledge discovery is the prime goal of the BRIEFS project.
The pilot domain for which the system is initially built is related to business intelligence and business planning. However, the system can be applied to other domains as well since there is nothing inherently tied to the domain of business.
One central idea behind the BRIEFS scheme is that the process of business intelligence in a particular domain of interest can be driven by a document, to be called here the Brief. The Brief describes the context in and for which information is desired. Ideally, the Brief gives, like for any other task to be undertaken, the definition of the task and sufficient background information needed for the successful execution of the task. E.g. if a technology domain is of interest the Brief could contain an expert state of the art description of the domain and its expected development and the current view about the significance (to the company) of the technology. If a business domain is of interest then the Brief should contain a description of the business and its future (the business plan!). The Brief thus is much more than a collection of simple key words or search phrases.
From the Brief it may be possible to infer what are the important concepts in the domain, what are their attributes and how closely the concepts may be related. (The choices made by the authors of the Brief also imply some tacit knowledge.) Based on the analysis a profile of the domain can then be built. The profile in turn serves as a reference base for evaluating the relevance of the acquired business intelligence to the particular domain.
The second important idea in the BRIEFS project is that associated with a domain of interest there can be some important hypotheses about the future, e.g. the commercialization of some technology by some date or the expected rate of development of the price/performance aspect of a technology etc. These hypotheses could be organized for semi-automated follow-up with the BRIEFS project. This would entail extracting domain knowledge from the Brief in order to set up a (object based) knowledge model of the domain. The hypotheses in turn define a template to drive the extracting of relevant information. From the documents that have been deemed relevant by the information retrieval process specific information could then be extracted with the extraction algorithms. These algorithms would find and extract desired data from the documents and fill in the pre-defined templates with the desired data. The idea about hypotheses has been treated in Keijola (1999).
The third important idea in the BRIEFS concept is a consequence of the mechanisms described above. When individuals or organisations learn to understand a particular domain better they may want to redefine their domain of interest and indicate change in their needs for information. Most of the work that is then needed to affect this change is that a new or modified Brief be written or that new or modified hypotheses be developed. In this way the system will satisfy the requirement for dynamism and adaptation, which is crucial in today's business environment. Knowledge will accumulate in the form of tested hypotheses.
A schematic outline of the BRIEFS system is in Figure 1-1. In the following sections we describe in more detail the BRIEFS prototype system that has been developed.
Figure 1-1. Outline of the BRIEFS 4 System.