products
Here you will find information about software products that parsQube GmbH has developed through extensive research.
We invest a significant amount of our time and effort in research and innovation. To this end, we set ourselves goals that break new ground in technology. The fact that these ideas are genuinely innovative is demonstrated by our research projects, which have been deemed eligible for funding by the Research Allowance Certification Office (BSFZ) since 2020.
The insights gained and the frameworks developed from these projects have also led to the creation of products that we have added to our portfolio.
On the following pages, we present the results of our research projects and the products that have emerged from them.
A comprehensive tool for document processing
Octopus is an innovative transformation engine that has emerged from many years of joint development work between parsQube GmbH and our sister company data2type GmbH. It consists of several components, offers a wide range of functions and is regarded as a comprehensive tool for document processing.
DigiSchool – (Word-based) digitalisation platform for schools
The demonstrator for a digitalisation platform developed by parsQube can automatically export learning content from Word into PDF and HTML formats. The innovative software engine requires no programming knowledge and significantly reduces the workload for teachers.
We initiated this project at the beginning of the COVID-19 pandemic. As parents, we experienced during 2020 and 2021 firsthand how difficult it was for schools and teachers to engage in any form of interaction with schoolchildren. At that time, homeschooling typically consisted of exchanging scanned book pages and assignments, which were distributed to students via Microsoft® Teams. The students would then print out the documents, complete the tasks, scan the pages again, and send the solutions back to the teachers via Teams.
With our project, we enabled several schools to convert assignments from Word to QTI or SCORM formats. This made real interactions with students possible without the need to invest in expensive licenses and training for tools like Moodle and similar platforms.
Digital time series manager for predictive maintenance management
parsQube has developed a digital time series manager for predictive maintenance management that can be used across a range of industries. These innovative methods enable predictive maintenance management even in sectors (such as chemical plants and laboratories) for which no standardized solutions have been available to date due to the complexity of the data involved.
Digital Case Assurance – Digitalised insurance cases
One of our clients, an insurance company, spent 1,5 years digitising their customer files. However, this digitisation process simply involved scanning the files – as images, which were then saved as PDFs.
As a result, staff were now working with PDF files instead of physical files. Furthermore, as the PDFs were saved under cryptic names, every file had to be opened when searching for specific information.
With this project, we explored how to process large volumes of PDF files using software that can recognise their structure and create a kind of database from the extracted data. This database enables almost any search query to be answered within seconds. The unique aspect of the results from this research project is not only the extraction of data and creation of a database, but also the ability to recognise the structure of the imported documents and filter out almost all metadata (such as the name of the doctor in examinations, his specialty, the date of the examination, etc.), which is of great value to the insurance company.
StrucRAG – Methodical framework for measuring and reducing information loss in structured RAG Pipelines
The goal of StrucRAG is the development of a methodical framework for measuring and reducing information loss that occurs during the processing of structured documents for RAG (Retrieval-Augmented Generation) systems. This includes the creation of a referenced corpus of structured documents, the development of suitable quality metrics as well as the design of layout- and semantics-aware representation and chunking methods. The results will serve as a scientific and technical foundation for future product developments.
