Open Password – Monday October 18, 2021
#987
Steep templates for corporate success – Art of decision – Oliver Berchtold – Best Practice – YUCCA Lab – Investment and risk decisions – Real-time news – Data analytics – Sentiment and event recognition – Media intelligence – Digital transformation – Business innovation – Design thinking – Information Overload – Filter Bubbles – Biases – Relevance -Start-up Scene – Berlin Ecosystem – Zurich Ecosystem – Natural Language Processing – Machine Learning – Artificial Intelligence – News Analytics & Prediction – Use Cases – Banks – Insurance Companies – Consulting Firms – Risk Lab – ESG Lab – Investment Lab – Globalization – Scaling – Corona crisis – Growth-driven investment strategy – Ratings – Aggregation Level – Reduction Level – Nowcasting Lebel – Forecasting Level – Automation Level – Risk Management – ESG Compliance & Reporting – Investment – Success Stories – Automatic recognition of SMEs – Expansion of Text recognition for additional languages
ZB MED – Medical Subject Headings – National Library of Medicine – FAIR Data Criteria – Open Science – DIMDI – TermCurator – KIBA – Virtual Education Fair – Further training and part-time offers – Elevator Pitches – Frauke Schade – Stefan Schmunk – German Research Center for Artificial Intelligence – BMBF – Anja Karliczek – Research Transfer – Spin-off Companies
- Cover story
Oliver Berchtold: Best practice at YUKKA Lab – Investment and risk decisions – Improve company performance with real-time news and data analytics
II.
E.g. MED: Medical Subject Headings in German
III.
KIBA: First virtual education fair for continuing education and part-time offerings
IV.
DFKI: Still of crucial importance for the implementation of the Federal Government’s AI strategy
Steep assists for corporate success in 2021
Online conference | October 20, 2021 | 10:00 a.m. – 1:30 p.m
THE ART OF DECISION
WELL INFORMED OR BETTER ADVICE?
Registrations: steep templates-2021.xcom.live
Oliver Berchtold:
Best practice in YUKKA Lab investment and risk decisions
Improve company performance with real-time news and data analytics
Oliver Berchtold, co-founder and head of product development at YUKKA Lab AG, will speak at the “Steilvorlagen 2021” on October 20th about “Better decision-making processes thanks to unstructured real-time data” (see Open Password, September 1st, #960).
The Berlin start-up is at the forefront of the development of context-based sentiment and event recognition based on natural language processing and has set out to convert unstructured data into indicators and signals for information experts in a wide range of industries and to take their processes and systems to the next level. Before founding YUKKA Lab, Oliver Berchtold led the DACH region’s market leader for media intelligence for over six years as head of product development through the digital transformation of the products and through the transformation of the development and decision-making processes towards greater agility. He put the knowledge he acquired during his studies about business innovation (University of St. Gallen) and design thinking (Stanford) into practice.
Oliver Berchtold
In your experience, how do companies and information professionals typically deal with “information overload” and what deficits need to be addressed above all?
In most companies there are no structured processes, guidelines or systems for how information professionals have to find out about the company, its competitors and the market. It is up to the information professional to decide whether he or she should inform themselves about relevant news and in what form. This represents a particular risk for the company. Because on the one hand we are all trapped in various filter bubbles and on the other hand we cannot get rid of several biases. This influences what news we see, what news we read and how we interpret it.
YUKKA Lab offers a data-driven process for selecting relevant news. Based on our indicators, the information professional can identify which companies are heavily exposed to which topics in the media and select the most relevant articles based on this preliminary analysis. This is not only more efficient, but the news is also searched for relevant developments in a much broader and more structured manner. This gives the information professional and his company a competitive advantage over his competitors.
What is your personal career, how did you and others come up with the product idea and then how did you found YUKKA Lab?
I grew up in Switzerland, studied business innovation at the University of St. Gallen and design thinking in Stanford. After completing my studies, I worked for the DACH region’s market leader for media intelligence for six years – first as head of strategic development, then as responsible for key clients and product management and finally as a member of the management team as head of product development.
The idea for founding YUKKA Lab goes back to the question of whether stock market prices can be predicted using news. This difficult and extremely attractive question prompted us to take it on as a business challenge. Today, hardly a day goes by when we don’t work with customers to question the status quo of a particular industry or process and consider how we can make the customer’s business faster, more efficient and better with the addition of our real-time news data. This is always incredibly interesting and great fun.
How important was and is the integration of YUKKA Lab into a start-up scene and the collaboration with science and partners? AI is evolving rapidly. How do you keep up with external developments?
Embedding yourself in the startup scene is fundamental for a start-up. With our locations in Berlin and Zurich, we were lucky enough to be integrated into two ecosystems. Berlin helps us find and recruit the “hot talents” for our mission and our company. As a hub, Zurich helped us to network with customers, as distances in Switzerland are shorter and the people in the ecosystem are extremely well connected with each other.
The majority of our development work takes place in the areas of Natural Language Processing (NLP) and Machine Learning (ML). This is where we have placed the focus of our investments and are in contact with several universities, for example the University of St. Gallen, the HTW Berlin, the University of Potsdam and the Saarland University. By exchanging ideas with other experts and science, we try to anticipate technological developments in the areas of NLP, ML and AI (artificial intelligence) and adopt the latest findings for ourselves. So far we have succeeded in doing this very well and have built an excellent reputation in the DACH region in the area of “News Analytics & Prediction”.
Important development steps of your company, for example in the context of financing rounds and where is YUKKA Lab economically today? In which sectors (e.g. finance) and regions (e.g. outside Germany) are you particularly successful?
YUKKA Lab started as one of the pioneers with the topic of “News Sentiment”. Start-ups in development-intensive areas such as NLP, ML and AI go through a longer initial development process. The AI needs to be developed, trained and optimized first if, like us, you rely on in-house development in these areas. After the initial phase, in which the technological feasibility must be proven, the aim is to develop the use cases together with customers. YUKKA Lab started with financial signals and then developed the SaaS solution that analyzes hundreds of thousands of news articles in real time and visualizes media trends based on them. When companies from other industries showed interest in our solutions, we expanded the customer portfolio from banks to insurance companies and consulting companies. To achieve this, we have differentiated our solutions according to target groups. Today, in addition to our News Lab, we offer a dedicated Risk Lab for risk managers, an ESG Lab for ESG Compliance & Reporting and an Investment Lab for Portfolio and Investment Managers (ESG = Environmental, Social and Corporate Governance).
The majority of our customers come from Germany and Switzerland. Since these are global companies, our orientation as a company and the further development of our products are becoming more global. In parallel to these developments, we are scaling our team and aiming to establish YUKKA Lab internationally through partnerships.
As an AI start-up, we pursue a growth-driven investment strategy. After Seed and Pre-Series A, we are financially well positioned. We are emerging from the Corona crisis stronger because customers’ need for indicators from unstructured real-time data increased sharply during the crisis, especially in the areas of risk management and ESG.
How do you want to achieve success in industries that are less technology-savvy or ratings-savvy?
News is highly relevant for all industries and business units. The market potential is therefore huge for us. It will always be about identifying, together with the customer, a meaningful application of our data and solutions that addresses and solves the problems of users in this industry.
We approach rating-related industries very specifically, since ratings depend on structured data and therefore, by definition, lag behind current events as manifested in the news. This offers us the opportunity to provide news data in real time as additional input for the rating and to offer our own ratings or forecasts on rating developments – as we do, for example, in the area of credit ratings for a leading global reinsurance company.
How can your product or product portfolio be summarized in a few words?
We transform text into data. Sentence by sentence, we analyze the semantics, the entities that occur, as well as the context and references. This enables us to quantify the media presence of every company, every industry and every topic. By examining over 700k articles daily from over 150k global sources, we measure the media landscape in real time. Based on this, we can identify trends and outliers and use events to determine the drivers behind the reporting. By combining them with other data sets such as stock market prices or credit ratings, we recognize correlations and patterns and derive predictive models from them.
It seems to me that important unique selling points of your product or product portfolio are the size and functionality of your archive and your ability to carry out real-time analysis.
That’s right. A central point, however, is our experience in developing indicators and forecast models for different domains. All of this depends on the quality of the underlying NLP and ML models, which is why we rely on our own developments and locate our IP here.
Let’s go through the stages of the YUKKA Lab AI concept for condensing news one by one.
The YUKKA Lab ” Five Levels of AI ” is structured as follows:
- In the first step, the ” aggregation level “, we aggregate all news about a company and enrich the articles in detail with metadata.
- , what is relevant for the user is filtered in the second level, the “ reduction level ”. With the reduction to level 2, the amount of information is reduced and focused to a level that the human brain can process.
- In the third level, the ” Nowcasting Level”, the current state for a specific point of view is reflected. Here we replicate our user’s point of view by creating a news-based score for areas such as risk management or ESG. This shows how high a company’s news-based risk is in the event of dangers such as money laundering or cyberattacks. The user can also build an individual score in which they select the relevant threats for their use case from our event universe, weight them and thus calibrate their own score.
- On the fourth level, the “ Forecasting Level ”, we answer the question about future developments based on the current status. Similar to the weather report with its five-day forecasts, we give short-term forecasts for credit ratings and investments, for example.
- The fifth and final level is the “ Automation Level ”. Based on the situation assessment using the scores from Level 3 and the forecasts from Level 4, sub-processes can be automated.
We have currently completed Level 3 for all of our solutions and are working on developing and establishing Level 4 for all use cases. How far we get with automation to Level 5 depends on the readiness of our customers. I expect more medium-term and long-term developments here.
Central application areas and examples from the perspective of the company and the individual knowledge worker?
– Risk Management : Structured screening of counterparties. Triage via risk score, whereby news-based risks are checked. Audit function, which allows critical messages to be forwarded to other departments for clarification such as compliance, including integrated audit track functionality. In addition, alerts for relevant changes in the risk score as well as forecasts for potential downgrades.
– ESG Compliance & Reporting : Monitoring and review of investments, supply chain partners and other relevant topics in the ESG area. Think of a pension fund manager who has invested the money in a variety of funds, which in turn hold a variety of stocks. He would like to know the exposure of all stocks to ESG topics in order to actively avoid reputational risks at an early stage and achieve good ESG performance.
– Investment: Combining sentiment with news-based ESG and risk scores to assess risks for fundamental or technical strategies. This means that customers not only achieved better but also more stable performance.
Do you have a success story for me about how a company became more efficient, more profitable or more innovative thanks to your engine?
Yes, for example a customer who reported to us a massive increase in efficiency by a factor of 25 through our ESG screening. The duration of the check per title, which had previously been carried out via a search engine, fell by a factor of 5. The number of titles to be checked also fell by a factor of 5, as our engine took over the preliminary check and then only companies with a risky score are checked had to. These results can be transferred to areas such as underwriting or sales.
For example, we received feedback from a B2B sales team at a telecommunications company that meeting preparation time could be reduced by up to 75%. The quality of customer discussions has also been greatly increased, as employees are now better and more widely informed in advance, for example with news leads about projects with existing customers and with hooks for contacting new customers. As a result, sales increased by more than 5%.
In the investment area, we reduce the maximum drawdown by up to 58% with our stock hedge signals. A company that integrated our news and news analytics into its customer-facing applications increased the time spent by its users by an average of 35%.
Technical and economic prospects of YUKKA Lab in the next two or three years?
The launch of the target group-specific platforms Risk Lab and ESG Lab by the end of the year will be an important milestone for the company and will allow us to respond even more closely to customer needs and address them more directly. When developing our products, we focus on the automatic recognition of small and medium-sized companies and the expansion of text recognition to include additional languages. Another focus of further development will be the expansion of our forecast models for the various areas.
As a company, we have just moved into new, larger offices in the Bikini House at the Berlin Zoo and are starting the process of scaling with the next Series A milestone. We will be putting more focus and resources into the areas of Marketing & Sales. We also give high priority to building partnerships and collaborations in order to further establish ourselves in several areas.
E.g. MED
Medical Subject Headings in German
The Medical Subject Headings – MeSH for short – are an internationally recognized and globally used biomedical thesaurus. The current edition of the German MeSH is now available for free download in various FAIR file formats. ZB MED – Information Center for Life Sciences created the translation for the German-speaking world for the first time. A specially developed semi-automatic translation process was used. The English-language original is published by the US National Library of Medicine (NLM).
In addition to the usual formats – i.e. XML and CSV – ZB MED offers the German MeSH for the first time in a bilingual German/English translation in semantically FAIR formats. These are, for example, RDF/XML or JSON-LD. This data therefore corresponds to the FAIR Data criteria – it is Findable, Accessible, Interoperable and Re-Usable. Specifically, for example, software solutions for data analysis that support the semantic web – including with artificial intelligence – can use the data directly because they do not need to be additionally converted and prepared. In the spirit of Open Science, the CC BY 4.0 license applies to use: Taking into account the terms of use, the reproduction and distribution as well as the modification and further processing of the German MeSH terms is expressly permitted and possible free of charge.
In 2020, ZB MED took over responsibility for the translation of the Medical Subject Headings from the German Institute for Medical Documentation and Information (DIMDI/BfArM). To achieve this, the team developed TermCurator, a semi-automatic translation tool with an integrated multi-stage curation process.
The MeSH thesaurus is one of the most important sources for a controlled biomedical vocabulary. In particular, he makes a contribution to categorizing and analyzing literature and data sources, for example in indexing media, indexing databases and creating search profiles. The MeSH is always up to date, as the NLM publishes an updated version every year. For the German-language version, the newly added terms are then translated and additional synonyms are added.
To download: https://www.zbmed.de/open-science/terminologien/deutscher-mesh/
KIBA
First virtual education fair
for continuing education and part-time offerings
(KIBA) Innovation and dynamics of change in technology and society lead to an increasing diversification of activities in libraries. New fields of action and challenges are emerging – for example in the areas of open science, digital humanities, research data management and the promotion of information, media and digital competence. In addition, the lack of skilled workers means that career changers have to qualify on the job.
How do facilities, personnel development and employees stay up to date and constantly develop their knowledge and skills in order to master current and future requirements? What part-time continuing education offerings are there and how do they respond to current and future developments and needs?
On Friday, January 14, 2022, from 4:00 p.m. to 7:30 p.m., the first virtual education fair for continuing education and part-time offerings in German-speaking countries will take place, organized by the Conference of Information and Library Science Training and Study Programs (KIBA), Section 7 is organized in the dbv and the DGI training commission. Presented in short elevator pitches
Further training opportunities and specific offers from information science institutes, departments and further training centers. All participants have the opportunity to find out about all the offers and ask personal questions in individual discussions. The program:
4:00 p.m. – 4:10 p.m.: Welcome Prof. Dr. Stefan Schmunk, Chairman of KIBA
4:10 p.m. – 5:00 p.m.: Elevator pitches
- Information Science BA, FH Graubünden, presented by Prof. Dr. Bernard Bekavac
- Information management part-time BA, Hannover University of Applied Sciences, presented by Anke Wittich
- Distance learning library science BA , FH Potsdam, presented by Prof. Dr. Ellen Euler
- Contact course in library and information management , HdM Stuttgart, presented by Prof. Cornelia Vonhoff
- Dual study program Information Science BA & MA , Darmstadt University of Applied Sciences, presented by Prof. Dr. Stefan Schmunk
- Information Science MA, FH Graubünden, presented by Prof. MSc, federal. dipl. Ivo Macek
- Library and Information Science MA , HU Berlin, presented by Dr. Ulla Wimmer
- Digital Data Management MA , FH Potsdam & HU Berlin, presented by Prof. Dr. Heike Neuroth and Prof. Dr. Vivien Petras
- Library and Information Science MA , TH Cologne, presented by Dorothee Heidebroek-Hofferberth
- Library Informatics MA , FH Wildau, presented by Dr. Frank Seeliger
- Library training , FU Berlin, presented by Christiane Preißler
- Part-time certificate courses, ZBIW Cologne, presented by Prof. Dr. Ursula Georgy and Franziska Weber
- University course Library and Information Studies, library training in Austria, presented by Dr. Gabriele Pum (Austrian National Library) and Mag. Alina Rezniczek (Vienna University Library)
- Certificate courses and university courses , library training in Austria, presented by Mag. Monika Schneider-Jakob MSc (University and State Library of Tyrol)
- Seminar program 4L – Lifelong Learning for Librarians , library training in Austria, presented by Mag. Birgit Hörzer MSc (Graz University Library)
5:00 p.m. – 5:10 p.m.: Break
5:10 p.m. – 6:00 p.m.: Individual advice
6:00 p.m. – 6:10 p.m.: Break
6:10 p.m. – 7:00 p.m.: Individual advice
7:30 p.m.: Conclusion
Participation in the online event is free of charge. Registration at: https://www.wit-wildau.de/zentrale-weiterbildungsmesse/
Frauke Schade and Stefan Schmunk, KIBA,
contact: stefan.schmunk@h-da.de , frauke.schade@haw-hamburg.de
German Research Center for Artificial Intelligence
Further crucial importance for the implementation of the Federal Government’s AI strategy
(BMBF) The Federal Ministry of Education and Research (BMBF) and the federal states of Berlin, Bremen, Hesse, Lower Saxony, Rhineland-Palatinate, Saarland and Schleswig-Holstein have signed a joint declaration of intent to further develop the German Research Center for Artificial Intelligence (DFKI). In it, the federal government and the seven states declare that they want to support the DFKI’s course as an intermediary between basic research and industrial research on artificial intelligence (AI) with a total of 22 million euros per year, half of which comes from the federal government.
Federal Research Minister Anja Karliczek explains: “Every investment in research into artificial intelligence is an investment in our future. Artificial intelligence is one of the crucial key technologies. With more than 600 highly qualified scientists, the DFKI is now one of Germany’s leading application-related AI research institutes with international appeal. I am very pleased that, together with the countries involved, we have succeeded in placing the DFKI on a solid, future-proof basis and in continuing to provide reliable financial support for AI research at the DFKI. Through the DFKI, our goal is to noticeably accelerate the transfer of promising AI research and development into innovative products, services and start-ups. The joint effort by the federal and state governments will help to further develop the DFKI strategically, in terms of content and quality. Now it is also the turn of the private shareholders of DFKI GmbH to also increase their contribution so that this important AI research center can continue its success story and make its contribution to the implementation of the AI strategy.”
By 2021, more than 140 former DFKI employees have been appointed to professorships worldwide. 98 spin-off companies were founded from the DFKI, of which 56 are currently still active. The DFKI plays a crucial role in the implementation of the Federal Government’s AI strategy. Many of the measures rely on the DFKI, with its experience, to be particularly involved in the transfer of research results into industrial practice, but also in the qualification of competent scientists. It is one of the research institutions that are to be strengthened as part of the federal government’s AI strategy. The prominent position of the DFKI is underlined by the fact that it is the only one of the German AI competence clusters that was successfully established long before the adoption of the AI strategy and is also the only institution to be financed with industrial participation.
OpenPassword
Forum and news
for the information industry
in German-speaking countries
New editions of Open Password appear three times a week.
If you would like to subscribe to the email service free of charge, please register at www.password-online.de.
The current edition of Open Password can be accessed immediately after it appears on the web. www.password-online.de/archiv. This also applies to all previously published editions.
International Cooperation Partner:
Outsell (London)
Business Industry Information Association/BIIA (Hong Kong)
Open Password Archive – Publications
OPEN PASSWORD ARCHIVE
DATA JOURNALISM
Handelsblatt’s Digital Reach