The specialist sits in front of the screen, swearing: A major customer has sent numerous delivery addresses in one order. When entering them, an employee has omitted the city. Another employee has mistakenly written the street name twice in the address field. Another colleague preferred to enter the delivery address in a comment field and added additional instructions. The assumption behind this: The recipient will find and correct everything. There are no limits to the creativity of the human mind, especially when time is of the essence and accuracy is a hindrance. Normal everyday life in a world based on the division of labor.

For this reason, human creativity should not be wasted on monotonous activities if these tasks can be performed automatically - and at lower cost and higher processing speed.

For the recipient of such electronic documents and data records, data entry is a nightmare. Regularly, address data in different places in a document and in varying quality means complete manual entry in the subsequent system. But that doesn't have to be the case!

| Heterogeneity in address data is manageable

A combination of intelligent text interpretation and a geocoding API such as Google Maps, Here, what3words or Geoapify can work wonders. Using proven RegEx methods, an address is divided into its components: Company, street, house number, city, zip code, country etc.. If it turns out that parts of the complete address are missing, these are automatically searched for and found using one of the aforementioned geocoding services via an API call. The check routine then inserts the missing parts so that the address is complete and correct.

This also works for different countries, as the RegEx patterns can be "injected" into the splitting algorithm, i.e. they can be dynamically introduced from outside the automation. This can also be used to log which addresses or documents contain the most errors, so that systematic improvements can be made by consulting the creators.

We proceed in a similar way when extracting address information from unstructured fields such as comments, free text or remarks. This is a particular challenge, as no commercially available service offers this. Many business partners write address information in any text field on the assumption that the recipient will recognize and use it thanks to human intelligence. However, this cannot be the goal. So we had to come up with a creative solution. Here, too, machine processing must take effect and recognize the unstructured address details in various free text fields and prepare them for further processing. For this purpose, we have developed a special algorithm that can be used in any business process automation. To validate the result, a geolocation API call can follow to ensure correctness.

| Free texts, comments and remarks are automatically readable

However, geocoding services offer even more interesting functions: For example, an address can be transmitted as a whole to Google Maps, what3words, Geoapify or Here. The response from the service called up then usually contains further information such as longitude and latitude or the type of address. This makes it possible to find out whether it is a building, a street and its characteristics, a restaurant or a specific type of business. This is relevant in cases where additional data such as transportation restrictions for trucks are useful for route planning.

Accuracy and completeness are a prerequisite for successful automation. Only reliable results are useful results. Otherwise, the manual effort required to correct incorrect data exceeds the benefits of automated processing. If implemented correctly, however, the desired quality of results can be achieved at a manageable cost.

Business Automatica reduces process costs by automating manual activities, increases the quality of data exchange in complex system architectures and connects on-premise systems with modern cloud and SaaS architectures.