BERD@NFDI

BERD@NFDI is an initiative to build a powerful platform for collecting, processing, analyzing, and preserving Business, Economic, and Related Data – all in one place. We will facilitate the integrated management of algorithms and data throughout the whole research cycle, with a special focus on unstructured (big) data such as video, image, audio, text, or mobile data.

BERD@NFDI will provide infrastructures for the challenges of the expanded empirical research. We will not only foster community building, offer publicly available and online accessible data sets, and enhance data documentation and preservation guided by the FAIR principles. We will also provide an algorithm repository and benchmarks, computing and storage power to analyze (big) data as well as a broad set of APIs to interact with external systems.

Learn more about us in our “BERD@NFDI in a nutshell“ presentation.

Our Mission

BERD@NFDI community involvement

The community of users representing the audience for BERD@NFDI infrastructure plays a key role in this project. We intend to establish a connection with the user community to keep it involved and part of the whole project. This connection enables a more active role of the users, starting from the conception and reaching to the design, implementation, and evaluation of the project. A user-centered design and an agile development methodolo- gy are an opportunity for an active and important role of the user community.

During the first phase, the user-centered design is used to identify the in – frastructure requirements from conception to implementation. The second phase continues to revolve around the user, but this time in a more “push” fashion. Now that infrastructure services are implemented, we will monitor usage activities, determine common patterns and make subtle, unobtru- sive suggestions to the user for maximum benefits from the infrastructure usage.

Measures:

Involving the scientific community and libraries
Methods of user-centered requirements engineering
Dissemination
Measuring impact

Collecting BERD

In social sciences in general and business and economics in particular, both the amount and type of data sources have proliferated. Sources of growing importance include unstructured data from social media, corporate reporting, communication of economic and governmental institutions and many more. Unlike with established data sources and common data types, there is a lack of an open, sufficiently documented and continuously improving process to adapt to novel and changing data source environments. Creating transparency of the types of data access and collection approaches will enhance the potential to reuse prior data collection efforts for replication, extension, and application to new research problems. BERD@NFDI aims to establish a process and rules to identify, evaluate and prioritize relevant sources of structured and unstructured data for social science research, particularly research in business and economics. Moreover, we plan to develop and establish processes and rules that enable researchers to evaluate and decide which data they are allowed to share with other researchers and provide the means to do so.

Measures:

Quality assurance and normalization
Anonymization
Processing Digitized Documents

Processing BERD

BERD@NFDI supports a community that works with many different resource types. Since they often do not stem from a single entity, they adhere to potentially different data management practices and can be created with different data collection approaches. Thus, researchers have to deal with a variety of data and different levels of data quality. If the data is already available in a structured form, established checks and normalization procedures can be used to improve the quality of the (meta-)data. For unstructured data, new methods of classification, normalization, and quality assessment have been applied, but there is no commonly accepted standard. When dealing with historical data sources, the printed sources must first be digitized using text recognition methods. In all cases, data protection requirements may make further processing necessary (e.g., anonymization to protect personal information). BERD@NFDI will support the research community in selecting suitable methods for processing BERD and in documenting and making them accessible. We will develop standards and guidelines for the processing and documentation of unstructured data, evaluate new anonymization methods and provide tools to manage the conversion of historical data sources.

Analyzing BERD

Researchers are not only searching for relevant data but also for relevant algorithms to be able to investigate novel topics. As analytical implementations and achieved performance are not consistently reported, it can be extremely challenging to choose a suitable approach for specific research questions, which severely inhibits scientific progress. To make data useful for research applications, BERD@NFDI will provide access to analytical capabilities that can be applied across various data sets. This will allow researchers to build on prior work, both in terms of data and analytical capability. It will also reduce redundant efforts across research units and address the demands of responsible AI. BERD@NFDI enables users to assess algorithmic performance both for data on BERD@NFDI and for their own data to quickly understand potential novel applications, as well as performance differences across applications of interest. Data storage, computing capacity, and analytical capabilities will be provided in one central infrastructure and will be complemented by algorithm repositories. Moreover, the infrastructure will support the enrichment and linking of data. Based on an ontology for business and economic data, a semantic network will be developed to capture complex relationships between companies and their social and historical environments.

Measures:

Data analysis tools
Semantic Enrichment and Linking of Data

Preservation of and Access to BERD

Access to a research data infrastructure and preservation of data provides the means to (re)use services for both human and machine scenarios. The information portal of BERD@NFDI will package all the necessary components in a delivery platform of preservation and access to users. It will provide search functionalities with different levels of granularity, a single sign-on solution in order to enable remote access to the infrastructure, and the persistent identification (PID) as well as migration of resources in BERD@NFDI. The prerequisite for the implementation of these functionalities is the development of a metadata schema that meets the requirements of the heterogeneous data in BERD@NFDI and is compatible with existing standards and approaches.

Measures:

FAIR Metadata Management
Search
Remote Access and Secure Data Storage
Information portal
Persistent identifiers
Migration Strategy

Re-using BERD

Simply providing a space for data sharing is not enough. We observe two barriers, unsolved by most current research data infrastructure facilities. First, researchers or agencies generating data often face resource or legal constraints when intending to share data. Most researchers (and often even agencies) do not have the time or money to prepare data in a format usable by others. This means they cannot provide support and context for those that are interested in analyzing the data, but are otherwise unfamiliar with the data generating process to do so properly. The convenient solution is to not share data in the first place. Second, even if the data is being shared, it often sits in archives with only a small number of users and little impact on academic insights. This is not because the data is not valuable, but because of insufficient support in understanding and using it, or difficulty in discovering it in the first place. BERD@NFDI aims to address both core barriers by creating an environment that demonstrates the value of sharing to those involved in data production. We will offer tailored training to the central target groups of BERD@NFDI in order to support them in (re)using the relevant data. Moreover, we will create the organizational, legal and technical conditions for potential users of the platform to be able to use the desired services without unnecessary barriers. Finally, we will create tools for automated feedback loops and data export functionalities to ensure maximum dissemination and discoverability of the data provided on BERD@NFDI.

Measures:

Education and Training
Data Stewardship services
Designing feedback loops: learning from past projects
Data export and integration with European Open Science Cloud (EOSC)

Research Data Management Strategy

Given the increasing relevance of unstructured and semi-structured data in empirical research in business, economics, and related fields, we propose to establish a holistic data management approach. The approach considers data management processes that are based on the UK Data Archive Lifecycle and governed by four major pillars, such as data governance (including data ethics), data integration and interoperability, data quality (including data cleansing, data integrity, data enrichment as well as data quality assurance) and data management infrastructure (including database and storage management).

About BERD

Our Mission

BERD@NFDI community involvement

BERD@NFDI community involvement

Measures:

Collecting BERD

Collecting BERD

Measures:

Processing BERD

Processing BERD

Analyzing BERD

Analyzing BERD

Measures:

Preservation of and Access to BERD

Preservation of and Access to BERD

Measures:

Re-using BERD

Re-using BERD

Measures:

Research Data Management Strategy

news

follow us