In the new normal, enterprises are focused more than ever on optimizing business processes, supporting employees, recognizing risk as early as possible, and shoring up resilience. Organizations are looking to get a head start in addressing these issues using artificial intelligence (AI), but there are other challenges to address, including complexity.
This is where academic and industry collaboration is essential, feeding off each other to overcome challenges, address real-world business cases and smooth the journey. For example, Professor Dirk Krechel and Professor Adrian Ulges and their team at the RheinMain University of Applied Sciences are exploiting a unique opportunity to work on scientific and technical problems in direct cooperation with industry to look at the business challenges of AI.
Getting under the hood of deep learning and content analytics
Currently, the team at RheinMain University is taking a deep dive into where there is a problem with AI approaches and how they can be solved. The “Deep Content Analytics” (DeepCA) project is a combination of deep learning and content analytics. The latter aims to harvest knowledge from heterogeneous data sources. In a business context, sources can be found in databases and applications.
DeepCA is funded by the Federal Ministry of Education and Research in Germany, and both sides — industry and academia — profit from this collaboration. Students learn about the real-world mandates and typical development processes found at a software vendor and at the same time, gain professional experience. In return, companies benefit from extensive academic research to support the development of next-generation software systems.
Within DeepCA, professors, students and researchers are primarily examining the AI algorithms which make it possible to look at data. In contrast to structured data, natural language texts cannot be searched and evaluated so quickly using conventional methods. This is where deep learning in the form of natural language processing (NLP) comes in and where intensive research is currently taking place. Academia aims to improve the search, classification, and categorization of documents for a corporate environment.
Why the need for analytics?
Why is it essential for enterprises to utilize technologies such as deep learning and content analytics? As the world becomes increasingly smarter, every enterprise needs a data and analytics strategy to deal with growing volumes of data and gain valuable insight and a competitive edge.
Enterprises are amassing vast amounts of data, but it is stored in an unstructured way, making it almost impossible to find centrally, let alone analyze for any real business value. Often this data isn’t adequately secured, leaving it wide open to data breaches.
As Professor Krechel explains, they have a large amount of data literally lying dormant. As a result, his team is exploring processes that allow data to be “easily tapped in an ECM environment and actively used in workflows.”
How does this work in practice?
There are classic extraction scenarios, for instance, where metadata is mined from unstructured information. Or there is semantic search, which enables companies to explore and tap into their internal document pools. The system can significantly improve the results of searches for similar documents relating to a specific business process, according to Professor Krechel.
Take searching legal documents as an example. Company lawyers often have to trawl through court judgments to find arguments to back a specific legal position to apply to their case. Search engines can help, but then the lawyer has to work through the documents. Paraphrasing, sentence structure and other factors all affect whether a passage from the text supports the argument. A similarity search allows texts to be found without the extra manual effort.
Part of DeepCA’s research revolves around investigating how the proximity of certain words to one another can be used to recognize similar documents. In a business context, this system allows users to search through existing contracts for specific clauses and retrieve all instances where the clauses have been superseded, making them invalid, for example.
So why aren’t more companies using AI in their processes? Professor Kerchel’s research has indicated that this is because of the amount of work involved in training AI models with the relevant data. Currently, a sufficiently large amount of data needs to be available for training to begin with. However, during his team’s research, they have shown limited sample data can still produce good results.
Models begin by learning from conventional research machines, such as using Okapi BM25, a ranking function used by search engines. The models can be refined using feedback from employees searching documents, for example. This approach allows enterprises to exploit the intelligence of significant search engines and adapt it to individual business needs.
Team members can define a business process via a small volume of documents and have the ECM system display similar processes from which they can take over responsibilities, for example. This saves substantial time on research and organizing workflows and provides intelligence for reliable decision-making, maintains Krechel.
Nurturing collaboration between universities and industry
Digital technologies and AI are creating enormous opportunities for industries. Collaboration between industry and academia is central to accelerating innovation and speeding up access.
Professor Krechel and his team are already working on follow-up applications for the project and plan to investigate concrete use cases for specific sectors, such as banking and insurance.
These two-way partnerships have much to offer. Industry and university collaborations such as DeepCA can help share knowledge and stimulate innovation, leading to accelerated advances in technology.