AMBALYTICS combines research results from various disciplines, including bibliometrics, mathematics and computer science, and aims to develop an easy-to-use web platform for the bibliometric workflow of knowledge workers.
This includes the following tasks:
- Exploration of new knowledge areas
- Targeted search for publications based on specific search criteria
- Acquaintance with the current state of research in a research area
- Supporting the authoring of own publications
- Support and monitoring of publication exploitations
- Comparison and connection with related research groups
- Recognition and analysis of research trends
Below you will find a selection of already implemented and planned functions that will support the above mentioned tasks.
The AMBALYTICS platform consists of a multitude of micro-services, such as publication search services, machine learning analysis services, distributed database and search clusters for bibliometric data, authentication services, and load balancers. Many of these services are stateless and can be easily scaled to provide, for example, AMBALYTICS services to a larger number of users.
The micro-services run containerized on a variety of hardware servers with GPU accelerators, which in turn form a Kubernetes cluster that can be quickly reconfigured and adapted to new circumstances using the Infrastructure-as-a-Service approach. Consequently, AMBALYTICS is equipped for future developments, e.g., for bibliometric analyses and machine learning.
AMBALYTICS uses a semantically annotated bibliometric graph stored and searchable in a high-performance distributed architecture. This is generated both by publicly available data, e.g. the Microsoft Academic Graph, and by applying proprietary machine learning algorithms. This makes it possible, for example, to determine the semantic context of two publications. Based on this, AMBALYTICS offers a variety of analyses for scientific publications.
In this way, the bibliometric structure of a research area can be displayed and analyzed. For example, thematically similar publications can be grouped into subject communities using hybrid algorithms that analyze citation data on the one hand and perform a content-based analysis of the publications on the other.
In a next step, it is planned to include information on research programs in the analyses, allowing, for example, the costs and benefits of a research funding program to be analyzed both for the funder and for the respective research institution with regard to scientific exploitation.
Furthermore, AMBALYTICS enables research trends to be identified based on calculated research fronts.
Clean User Interfaces
The most advanced technologies are of the least use if they are not usable. The algorithms used by AMBALYTICS services are also complicated to parameterize and implement.
The AMBALYTICS platform has been consistently designed from the very beginning to provide a simple and friendly user experience. Thus, cluster analyses of publications can be performed with "one click" and do not require any learning on algorithms or programming.
Furthermore, the AMBALYTICS platform is designed to support the workflow of scientists in the best possible way. For this purpose, in addition to the innovative analyses, we are creating functions such as storing, sorting, analyzing and sharing of publications, collaborative functions, seamless interfaces with other tools (such as for Mendeley, Zotero or Overleaf), automatic reports generation, or notifications, which facilitate the day-to-day handling of scientific publications. A mobile companion app is in preparation as well, which will allow the AMBALYTICS platform to be used on the go.
Searching for publications with classical publication search engines, such as PubMed or Google Scholar, require the review of hundreds of search results in tabular form.
The core of AMBALYTICS is an easy-to-use and configurable graph representation. Each publication is represented as a node in a graph, and edges that can be shown or hidden represent direct citations or thematic connections. This allows a quick visual overview of a subject area and a faster detection of relevant publications.
Previous and follow-up publications can be discovered for a specific publication as well as related publications can be found more quickly.
In addition, the graph can be customized for different aspects, for example, to highlight trends or changes over different time periods, or thematic similarities.
We are also developing a 3D application that shows clustered bibliometric entities, such as publications and authors, in a huge 3D space to enable exploratory and interactive inspection of large volumes of publications. The following screenshot contains 100.000 publications and is completely and interactively rendered in a browser.