Technology

Technology

The Challenge

The initial idea for kiresult came out of a project where we had to consolidate, cleanse and analyse spend data. Required parameters (e.g. category of goods and services) could be derived from order texts. But this took an enormous manual effort and a lot of time, which we did not have either. A better way had to be found.

Order and invoice texts are sequences of words that are interpreted by vocabulary we are familiar with. Understanding the meaning in the respective context is also important. For example, it is common practice for buyers to classify order texts and assign them to categories.

If this understanding could be transferred to a computer, assignments could be made automatically. This task pushes classical methodologies to their limits.

The Challenge

The initial idea for kiresult came out of a project where we had to consolidate, cleanse and analyse spend data. Required parameters (e.g. category of goods and services) could be derived from order texts. But this took an enormous manual effort and a lot of time, which we did not have either. A better way had to be found.

Order and invoice texts are sequences of words that are interpreted by vocabulary we are familiar with. Understanding the meaning in the respective context is also important. For example, it is common practice for buyers to classify order texts and assign them to categories.

If this understanding could be transferred to a computer, assignments could be made automatically. This task pushes classical methodologies to their limits.

Technological solution behind kiresult

The solutions of kiresult are based on machine learning, more precisely the field of Natural Language Processing (NLP). Using historical data, such as orders with category information, the computer learns the underlying rules, from the structure of used vocabulary to the interpretation of words.

kiresult works where  knowledge and experience of people is required.

Order texts are assigned to categories of goods and other attributes, such as order value or supplier. Kiresult transforms all this information into a specially developed data model to learn and remember the rules and individual correlations contained. The models can then be applied to all future data sets.

Technological solution behind kiresult

The solutions of kiresult are based on machine learning, more precisely the field of Natural Language Processing (NLP). Using historical data, such as orders with category information, the computer learns the underlying rules, from the structure of used vocabulary to the interpretation of words.

kiresult works where  knowledge and experience of people is required.

Order texts are assigned to categories of goods and other attributes, such as order value or supplier. Kiresult transforms all this information into a specially developed data model to learn and remember the rules and individual correlations contained. The models can then be applied to all future data sets.

The speciality of kiresult

Processing of individual data sets requires consideration of a wide variety of conditions. This implies a precise alignment of used algorithms. The software of kiresult can flexibly cover all these aspects. This enables to develop solutions for our customers fast, to provide them permanently and to make them usable.

Besides the solution of the problem, knowledge transfer has top priority.

Finally, not only the software, but also the team is what makes kiresult stand out. The entire solution path is closely tracked by open communication.

The speciality of kiresult

Processing of individual data sets requires consideration of a wide variety of conditions. This implies a precise alignment of used algorithms. The software of kiresult can flexibly cover all these aspects. This enables to develop solutions for our customers fast, to provide them permanently and to make them usable.

Besides the solution of the problem, knowledge transfer has top priority.

Finally, not only the software, but also the team is what makes kiresult stand out. The entire solution path is closely tracked by open communication.

Do you want to unlock your data potential?

Receive further information by e-mail.

We offer a free feasibility check, where we identify your data potential and develop initial ideas.

  • Initial call

    In a first exchange we discuss your needs. We will contact you to find a suitable time.

  • Analysis

    We check the quality of your data structures and derive suitable use cases for our algorithms.

  • First results

    We present your identified potentials. Then we define next steps for implementation.

Want to unlock your data potential?

Receive further information by e-mail.

We offer a free feasibility check, where we identify your data potential and develop initial ideas.
  • Initial call

    In a first exchange we discuss your needs. We will contact you to find a suitable time.

  • Analysis

    We check the quality of your data structures and derive suitable use cases for our algorithms.

  • First results

    We present your identified potentials. Then we define next steps for implementation.