Business intelligence (BI) is a term used to describe the strategies and technologies used by companies to analyse data from business information. BI tools offer historical, current and prognostic perspectives on business operations – with the aim of making business decisions based on data.
Sources for business analytics can be structured data from internal reports, a merchandise management system, CRM or ERP applications and other systems. But external sources such as third-party providers or webtracking services also provide important input. More difficult to include in business intelligence are the unstructured data generated in e-mails, presentations, memos and other corporate content.
In concrete terms, business intelligence can be used to answer questions such as: which products or services deliver the best ROI for which marketing campaigns? Are there particularly high return rates in defined sales regions within certain periods of time? Where is the potential for expansion due to good economic conditions?
With big data, the challenge increases to filter meaningful information from the jumble of numbers, to prepare it understandably and to interpret it in a target-oriented way. The main task of a series of tools from the area of business intelligence is therefore the aggregation and visualisation of business-relevant data. Business analysts use such applications to find, analyse and present data in line with the company’s strategic goals.
For the correct operation of such tools and an understanding of the information gained, sound analytical skills are usually a prerequisite. Even at management level, explanations are occasionally required on how to classify complex tables and charts in a targeted manner.
Where companies today are under pressure to ensure that employees in all departments gain a transparent insight into developments and make secure decisions for their area, so much potential is wasted.
Natural language processing (NLP) can make an important contribution to understanding business intelligence. Simply explained, natural language processing describes techniques and methods for the computer-aided processing of natural language to facilitate direct communication between humans and machines. NLP is part of the large field of artificial intelligence and is divided into the areas of natural language understanding and natural language generation.
The link between business intelligence and natural language processing has two meaningful dimensions. On the one hand, natural language understanding (NLU) can be used to convert unstructured company content into structured data. NLU applications “understand” relevant information from a stock of existing documents, filter out raw data and thus make it accessible for further processing in BI analysis tools.
On another level of business intelligence, natural language generation creates descriptive and interpretive reports from data. If, for example, data shows the increase of a value, automatic text generation adds a descriptive text to the corresponding diagram.
Beyond this, natural language generation is also able to serve other aspects of a text.
A mere description of data often leaves a user at a point from which he or she does not know where to start. More interesting, then, is the classification of a phenomenon. For example: is a red number in a segment a serious incident or perhaps just part of a recurring problem? Such correlations can be demonstrated in text form through the interaction of business intelligence and natural language generation.
Business intelligence is not an end in itself. Rather, the analysis is about concrete conclusions for one’s own area of responsibility. Natural language generation has the potential to activate users via language. If practical instructions result from the numbers, automated text generation can formulate these for the user.
The great advantage of natural language processing in business intelligence becomes clear here: users do not have to be proven business analysts to recognise correlations and draw conclusions. Regardless of department, management level, prior knowledge or geographical location, data and the corresponding interpretations become accessible to everyone. In this way, natural language processing democratises business intelligence.