Biological and medical researchers explore the mechanisms of living organisms and tend to gain a better understanding of underlying fundamental biological processes of life. To tackle such complex tasks they constantly need to gather and accumulate new knowledge by performing experiments and studying scienti c literature. We will present the novel semantic search engine "SCAIView" for knowledge discovery and retrieval and, additionally, discuss the most recent paradigm shifts in communication technologies, which leads to a completely new architecture that improves scalability, achieves better interoperability, and also increases fault-tolerance.
Biological and medical researchers are interested in exploring the mechanisms of living organisms and gaining a better understanding of underlying fundamental biological processes of life. To tackle such complex tasks they constantly gather and accumulate new knowledge by performing experiments, and also studying scienti c literature that includes results of further experiments performed by researchers. Existing solutions are mainly based on the methods of biomedical text mining to extract key information from unstructured biomedical text (such as publications, patents, and electronic health records).
Especially in the eld of biomedical sciences, we have a long history of
developing applications that solve the above mentioned tasks. For instance,
SCAIView3 is an information retrieval system that allows semantic searches
in large textual collections by combining free text searches with the ontological
representations of automatic recognized biological entities (see Hodapp et al.
[
http://academia.scaiview.com/academia/)
version is freely
available
at
real-world task is the creation of biological knowledge graphs that is tackled by
the BELIEF environment [
In this poster we will present the recent development of SCAIView, and how SCAIView (as well as BELIEF) evolved using the same core technologies to an interoperable software system. 2
To keep up with the state-of-the-art technologies and to be prepared for
integration of novel and game-changing developments, we migrated the SCAIView
ecosystem from a large monolith to microservice-based system. It allows us to
reuse parts for di erent purposes and the data itself can be easily processed,
shared and accessed. Additionally, the new system also allows us to focus on
FAIR (Findable, Accessible, Interoperable, and Reusable) principles, introduced
in [
The microservice infrastructure of SCAIView is an ecosystem of three main
services: Core, API, and Indexer (Figure 1), which communicate through the
message broker (Apache ActiveMQ). The core ful lls various important tasks
to persist, retrieve, and process data. Beside further text mining microservices,
there are also specialized microservices such as BEL Commons Professional,
which allows to validate text-mined biological entities and relationships, that are
shared by BELIEF and SCAIView ecosystems. SCAIView's user interface itself is
a web-based microservice application running on Apache Tomcat communicating
via REST-API calls with the backend. The visualization of the document corpus
includes document elements that are stored and represented as semantic digital
assets (SDA) (Jacobs et al. [
The Document Store is based on Apache Accumulo and Apache Solr. The
rst one is used to persist raw results of the text mining pipelines. This allows us
to compare and validate the development of old and new text mining components
really fast, which is necessary in the research area. The latter one contains SDAs
such as the document text, recognized semantic concepts, and further metadata
that is needed for fast retrieval. SCAIView can also handle multiple text
mining and knowledge discovery pipelines by communicating through the message
broker. Common steps are the usage of a DocumentDecomposer, Lemmatizer,
JProMiner for named entity recognition. Other text processing components, such
as UIMA Ruta-based components (see [
Search queries and knowledge discovery in SCAIView is linked to ontology
and terminology data. Semantic searches are a combination of free text search
and entities represented in ontologies or terminologies. For instance, SCAIView
includes Alzheimer's Disease Ontology (ADO), BioMarker terminology, drug
names, the Hypothesis Finder and many more. These resources are displayed
in a tree format and can be used to make detailed, faceted search queries and
to perform statistical analysis on the retrieved document corpus. The access to
these resources is provided by our internal-hosted OLS service (Ontology Lookup
Service [
In general, SCAIView is developed to handle any kind of document corpus but
currently we focus on the biomedical research area. Therefore, as input we use
databases such as PubMed 2017 [
Although several risks and problems have to be faced, we are sure that positive advantages of implementation of a microservice system do outweigh. For both applications, SCAIView as well as BELIEF, several microservices are used and shared for purpose of data retrieval, data persistence, and text mining. The latter are classical microservices, whereas the retrieval and persistence services are more general microservices. Additionally, the microservices in the data layer can also be traditional webservices such as the terminology management or authentication systems. We bene t from a highly scalable and fault-tolerant environment for data processing. Furthermore, the system is exible enough to easily add or remove microservices from the processing pipeline. The continuous
delivery process for externally-developed software like OLS or Keycloak is not an issue anymore. An additional bene t is the safe and fast switching from one technology to another: TeMOWl and OLS can be used at the same time for multiple instances of SCAIView.