Automatic document classification

Challenge

BSH needed to streamline and automate its analysis of after-sales technical services. These services are very often performed manually by the Quality Management Department.

This caused, for example, in the Esquíroz factory (Navarra):

• 10 days of work per month by an engineer from the department to collect data from repair reports previously filled in by technicians, classify them and cross-check them with production data to analyze quality indexes.

• the impossibility of manually classifying and analyzing all the technical services of the previous month

Solution

To address this challenge, PredictLand has designed two Artificial Intelligence solutions to automate the classification of Technical Repair Service parts.

The solution, which uses advanced Deep Learning and Natural Language processing techniques, interprets the free text indicated by the repair technician in his own words and language. With this interpretation, it is able to automatically identify and classify the service by type of fault, drastically reducing the engineers’ workload from 10 days a month to 6 minutes.

The solution was initially developed in one of the factories, and after its success it has been replicated in all factories in Spain.

Testimonials

Unai Esparza Baquer
Head of Market Quality at BSH Electrodomésticos, Esquiroz Factory (Navarra, Spain)

“In order to implement at BSH the ARIS project for automatic classification of repair services and market quality analysis, two years ago we decided to turn to PredictLand.

Now that ARIS is operational, having achieved its objectives and garnered awards such as the BSH Innovation Award 2020 and the Excellence Award 2020 from the German Chamber of Commerce, we can see that the decision to embark PredictLand on this adventure was the right one.

Without their extensive technical knowledge of Data Science we would not have been able to do it. In addition, we are grateful for the high degree of customer orientation shown by PredictLand in making an effort to know in depth our internal jargon and the resources we have at BSH, to explore the work strategies that best fit the internal functioning of our company and to offer training support in Data Science to BSH staff involved in the project.

It should also be noted that the team has worked in a permanent climate of trust based on transparency in terms of project feasibility analysis, cost estimation, risk identification and detailed and periodic follow-up of executed and pending tasks.

In this time we have created a solid and efficient team that extends beyond our company and from which I expect great success in the future as well.”

---

Francisco Javier Alemán Ezcaray
Quality Manager of the Refrigerator and Dishwasher Plants at BSH Electrodomésticos, Esquíroz Factory (Navarra, Spain)

“In 2018, the market quality manager of my department came to my site to talk to me about the infinite possibilities that a Machine Learning algorithm could give us when it came to analyzing the data that the technical service collected on the incidents of our devices in the market. At that moment, the idea was only in his head and that of Javier Orús, from PredictLand. The spark had been ignited in a training given by Javier himself and attended by people from my team.

We got down to work, in a very close collaboration between BSH and PredictLand, where we brought our knowledge of our market and devices, as well as our needs, and PredictLand a tremendous expertise and a very high understanding of the challenge we had set them.

As a result, we obtained a project that makes us have much more accurate data than we had before, in much less time and easily editable and understandable by other areas of BSH that until then had not dived into what was happening to our appliances in the market.

The qualitative leap in our analysis is not measurable. The progress is so great that I cannot compare it with the previous situation. And all of this is thanks to the great professionals at PredictLand, where not only their knowledge has been key, but also their attitude, closeness and constant support.

We already have a battery of projects in mind arising from how much we have learned with this project about the infinite possibilities of artificial intelligences.”

---

Borja Lara López
Industry 4.0 Coordinator at BSH Electrodomésticos España.

“During the time I have shared with PredictLand and its members in the development and execution of a project of high value for my company, I have had the privilege of working with people of a high technical capacity who have not only contributed very positively in the good development of the activities but have also contributed a lot of knowledge to the people who have participated in it.

Not only is the aptitude of this group of professionals good, but their attitude is also exemplary. During the course of the project, we encountered multiple pitfalls that were addressed with a very intelligent strategy, understanding the reality of our company and above all and very importantly with a magnificent flexibility to adopt other ways of working or tools that until then were alien to them.

It has not only been an enriching experience on a technical level, but also on a personal level, which has not failed to add value to us in multiple fields.”