Skip links

PoC for Comparis: Analyzing unstructured documents using AI.


BackgroundDeep Impact and Comparis have successfully structured unstructured data from Harvard Health publications using AI.

By using AI and innovative algorithms, the amount of data was automatically grouped into thematic clusters and a navigation hierarchy was created. This enables Comparis users to search and use the information much more efficiently.



Work deliverables



AI, Consulting

The challenge

Comparis was faced with the major task of structuring a large amount of unstructured data from Harvard Health publications. The aim was to prepare this data in such a way that it was easily accessible and usable for users in order to fundamentally improve the quality of service and the user experience on the platform.

The innovative solution approach of Deep Impact

Deep Impact met this challenge with an innovative approach that involved the use of artificial intelligence. An important step in this process was the development of proprietary Doc2Vec and Word2Vec algorithms by the Deep Impact team.

These algorithms are the result of intensive research and development and enabled a customized solution for Comparis’ specific requirements and data sets.

Der Proof of Concept (PoC)

For the PoC, Deep Impact used 3,500 documents from Harvard Health publications to demonstrate the AI’s ability to structure and group this data. Heliospan, an advanced product from Deep Impact’s subsidiary AVA-X, was used.

It was specially developed for processing and analyzing large volumes of unstructured data and made it possible to make the data clear and accessible.

Implementation and results

Heliospan’s AI engine automatically grouped the data into thematic clusters and created a navigation hierarchy. This automatic structuring made the data not only clear, but also easily accessible for Comparis users.

To validate the efficiency and accuracy of the AI engine, it was additionally tested with all the material from Wikipedia, successfully clustering and structuring data even without specific expert knowledge from the healthcare sector.

Conclusion and outlook

The case study with Comparis underlines the power of AI in modern data processing and positions Deep Impact as a technology leader in this field.

The independent development and successful implementation of complex algorithms by Deep Impact demonstrates a significant competitive advantage and opens up perspectives for future innovations and developments in the digital landscape. The successful PoC with Comparis shows the enormous potential of AI-supported data structuring, which enables companies to improve the accessibility and usability of their data. Deep Impact plans to further develop this technology and apply it to other industries and challenges to drive digital transformation and enable a better user experience and competitive advantage through the use of advanced technologies.

Further Information:

Deep Impact AG:

Share on