naturalAll kind of industries generate and collect an unimaginable quantity of data nowadays. According to an IDC report, the amount of data will grow globally from 33 zettabytes (33 trillion gigabytes) in 2018 to 175 zettabytes in 2025. Unfortunately, even large companies with huge budgets and big teams are already facing many unresolved challenges regarding the processing, evaluation, and efficient use of these huge amounts of data. The US analysis and consulting company Gartner, well- known for its in-depth studies such as the "Magic Quadrant" or technology forecasts like the "Hype Cycles", provides some seminal solutions for these challenges and ‘Natural Language Processing’ (NLP) and ‘Conversational Analytics’, our topics of the month, play an important part here.

Natural Language Processing (NLP) is a branch of linguistics and computer science that deals with the interactions between computers and human (natural) languages. Currently, companies are particularly concerned with the question of how to program computers to process and analyze large amounts of natural language data. This applies to search engines, voice commerce, voice assistants as well as analytics applications, especially since Gartner claims that 50% of analytics queries will be generated in systems via search, voice input (NLP) or automatically this year. The end customer trend of voice control in the car, for instance, via smartphone or smart speaker, is indeed getting more and more popular in B2B analytics applications. Hence, it is predicted that already by next year, processing natural language will increase acceptance of analysis and business intelligence software from 35% of employees to over 50%. This will make analytics more usable for new user groups such as managers, salespeople, or creative people. Such NLP functionalities are offered, for instance, already by companies like Qlik or Tableau.

Commercial AI & ML

Currently, many popular Artificial Intelligence (AI) and Machine Learning (ML) software frameworks are still supported by open source (e.g. TensorFlow from Google or Caffe by Berkeley AI Research). By 2022, 75% of new end user software (e.g. apps & websites) that use AI and ML technologies will work with commercial rather than open source software. Hence, Gartner forecasts that commercial cloud-based services from major providers (especially Amazon, Google, or Microsoft) will reach the turning point of 20% market share in the data science platform market by 2022. Especially since these large tech companies have long recognized the potential of data science and started to work on the commercialization of their self-developed frameworks early.

Explainable AI & Continuous Intelligence

Gartner expects more than 75% of large companies hiring their own AI specialists in areas such as IT forensics or data protection by 2023 to reduce risks for the reputation of their brand. Automatically generated insights and models are increasingly used with the help of augmented analytics. However, the explainability of these insights and models (e.g. their derivation) is crucial for trust, compliance with legal regulations, in other worlds, the management of the brand reputation. The fact is that inexplicable decisions, which are made by algorithms, often do not trigger enthusiasm in most people: some AI applications are said to reinforce certain prejudices or "learn" from training data. Explainable AI is a model whose strengths and weaknesses can be identified. The likely behavior of such a model can be predicted as well as possible distortions. An explainable AI makes decisions of a descriptive, predictive, or prescriptive model more transparent. In this way, important factors such as the accuracy, fairness, stability, or transparency of algorithmic decision-making can be ensured. By 2022, more than half of all major new business systems will have continuous intelligence that uses real-time context data to improve decisions. Continuous Intelligence combines raw data and analysis with transactional business processes and other real-time interactions. Methods such as event stream processing (a method for real-time analysis), business rule management (rule-based decision systems) and of course machine learning are used for this. Continuous intelligence can also be described as a further development of operational intelligence.

Augmented Analytics & Data Management

In general, data analysis is complex and requires one or more data scientists who can extract value from large amounts of data. The complexity is mostly because data is collected from different sources such as web analysis, enterprise resource planning (ERP), product information management (PIM), marketing software or social media. Due to the high manual effort for the preparation, cleaning and merging of data, data scientists spend most of their time with such tasks, which is estimated to be up to 80%. Augmented analytics can help here to reduce workload with machine learning that enables data scientists to invest more work in the search for actionable insights. By 2020, Gartner expects augmented analytics to be a dominant driver for business intelligence purchasing decisions, as well as data science and machine learning platforms. Augmented data management can help reduce the manual effort described above by cleaning and merging large amounts of data from different sources with machine learning and automated workflows.

Blockchain in Analytics & Graph Analytics

The promise of the blockchain is gigantic: it contains cryptographically protected data that cannot be changed. The data can only be shared and supplemented by a network of participants, so that everyone always has the latest data/information at the same time. Of course, this is extremely complex. Even so, blockchain technology has long been considered one of the technologies that will revolutionize retail. Analytics use cases would be, for example, fraud analysis, auditing processes or data sharing between different organizations. However, due to the blockchain-based applications that have so far not been convincing, many experts label this trend as hype. Well, we will see how it develops over the long term. Gartner however predicts, the use of analytical graphics processing and graphics databases to grow by 100% annually in the next two years, which allows carrying out complex data science much faster. A special application of Graph Analytics (graphic-capable semantic knowledge graphs. like the well-known ‘Google Knowledge Graph’) forms the basis for many natural language processing / conversational analytics data structures and enormously enriches many data processes. Generally, graph analytics describe several analytical techniques that make it possible to investigate relationships between organizations, people, and transactions.

By Daniela La Marca