Written by missing user171 do not delete on 24 March 2025. Posted in CASE STUDY

How the Sloane Lab, a collaboration between major universities and museums, unified its massive historical and archival collection using knowledge graphs with metaphactory

 

Bridging the past and present: Unifying the Sloane Collection with knowledge graphs

 

About the Sloane Lab

 

Sloane Lab is a research project funded by UK Research and Innovation (UKRI) as part of the “Towards a National Collection” program. The initiative explores technologies and methodologies for creating a unified national collection for the UK, ensuring that historical archives and artifacts are accessible and comprehensively documented. The project is a collaboration between University College London (UCL), Technische Universität Darmstadt (TU Darmstadt), the British Museum and the Natural History Museum London, with additional involvement from the British Library. The project was led by Prof. Dr. Julianne Nyhan, Professor of Humanities Data Science and Methodology at TU Darmstadt and Emeritus Professor of Digital Humanities at UCL. Dr. Andreas Vlachidis, Associate Professor of Information Science at UCL and whom we spoke to about the project, acted as project Co-Investigator and Technical Lead.

 

 

The challenge: Reconnecting a fragmented collection

 

The Sloane Collection, originally compiled by Sir Hans Sloane in the 18th century via a network of traders, donors and enslaved individuals, is a vast and meticulously cataloged collection comprising 50,000 books and manuscripts, hundreds of thousands of natural history specimens, hundreds of paintings and albums of prints and drawings, and thousands of ethnographic objects and antiquities. The collection was initially housed by the British Museum, but as the museum expanded over time, items from this collection were distributed across multiple institutions, including the Natural History Museum and the British Library. Each institution adopted distinct systems for managing metadata, leading to significant fragmentation of the collection and making it difficult for researchers to ask questions across the collection as a whole.

 

In a digitally reunified form, the Sloane Collection could better support research into the legacies of transatlantic enslavement, empire and colonialism and heritage collections. The Sloane Lab project aimed at bringing together historical and present-day information that describes the Sloane Collection and making this information available for exploration. Therefore, the primary challenge that the project was looking to tackle was the aggregation and unification of the collection’s metadata by integrating data from original manuscripts (digitized and transcribed catalogs) with modern museum catalogue systems while maintaining historical authenticity. Additionally, the project sought to address gaps within the historical records – such as the limited representation of women, enslaved people and indigenous communities – by developing a model capable of capturing such absences.

 

 

The solution: A knowledge graph for seamless integration and discovery

 

From the outset, the project adopted a knowledge graph approach to seamlessly integrate data into a unified Sloane Lab Knowledge Base and semantically enrich it. Beyond this, the project consortium envisioned an intuitive application that would empower non-expert technical users to engage with the compiled knowledge base. The goal was to transform exploration into an intuitive and dynamic experience, enabling users to uncover connections and make discoveries with ease.

 

To achieve this, Sloane Lab leveraged Ontotext GraphDB as its triple store and developed an ontology based on CIDOC CRM, extending it to model the semantics of absences within the collection. This ontology was then ingested into metaphactory, where it served as the foundation for an interactive application designed to facilitate querying, exploration and visualization of the knowledge base in a user-friendly and insightful manner.

 

 

Why metaphactory: Enabling intuitive knowledge exploration and discovery

 

In selecting a platform for making the Sloane Lab Knowledge Base accessible, the goal was to ensure that the knowledge graph was not only a powerful data integration tool but also an accessible and intuitive resource for both researchers and non-technical users. metaphactory was chosen for its ability to bridge the gap between complex semantic technologies and user-friendly interaction, offering:

 

  • A user-friendly search, discovery and exploration interface that makes the knowledge graph easily accessible to researchers and non-technical users alike.

  • A model-driven app building approach and template framework that simplifies the creation of custom views and interfaces, enabling flexible application development on top of the knowledge graph.

  • Comprehensive querying and visualization capabilities that provide rich insights and intuitive exploration over an extensive information layer.

 

metaphactory-powered visualizations from the Sloane Lab Knowledge Base



Notably, the application building was conducted entirely by the Sloane Lab project team led by Dr. Andreas Vlachidis, demonstrating the ease of use and adaptability of metaphactory. "metaphactory enabled our team to rapidly build an intuitive and flexible application on top of the knowledge graph without requiring extensive front-end development. Its model-driven approach allowed us to focus on enriching and structuring the knowledge base, while its powerful visualization and search capabilities made the collection accessible to a much broader audience," commented Vlachidis.

 

Key benefits: Transforming metadata into meaningful insights

 

The Sloane Lab Knowledge Base transforms the way the Sloane Collection is explored and understood by unifying fragmented metadata, enhancing accessibility, preserving historical integrity and establishing a scalable foundation for future expansion. Specifically, it delivers concrete benefits such as:

 

  • Unified collection view: The project successfully aggregated and connected disparate metadata from multiple institutions, bridging historical and contemporary information describing the Sloane Collection.

  • Enhanced accessibility: By using metaphactory, researchers and historians gained an intuitive and interactive platform for exploring the knowledge graph, reducing reliance on technical expertise.

  • Historical integrity & insights: The ontology captured the original composition of the collection, while also addressing historical biases and absences, enriching the understanding of its provenance.

  • Rapid application building: metaphactory’s flexible model-driven app building and templating framework allowed the team to easily create customized views and interfaces tailored to the project’s needs.

  • Scalability & future-proofing: The knowledge graph infrastructure established a scalable foundation for future expansions, enabling integration with additional datasets and institutions.

 

Through the innovative use of knowledge graph technology and metaphactory’s intuitive knowledge discovery interfaces and app building framework, the Sloane Lab project demonstrated how digital tools can reconnect fragmented historical collections, offering scholars and the public a more comprehensive and nuanced understanding of history.

 

 

Explore the Sloane Lab Knowledge Base

 

The Sloane Lab Knowledge Base provides an interactive platform for discovering and exploring the historical and current artifacts and records of the Sloane Collection. To browse interconnected data, visualize relationships and gain deeper insights into the collection’s history, start exploring here!


 

 

Written by missing user171 do not delete on 22 July 2024. Posted in CASE STUDY

How Ontotext leveraged metaphactory to develop its new risk and event detection platform.

 

Detecting & tracking market-moving events with a metaphactory-powered platform 

 

The customer

 

Ontotext is a knowledge graph software company offering an RDF database called GraphDB, a highly efficient and robust graph database with RDF and SPARQL support, compliant with W3C standards, that can help build big knowledge graphs.

 

The team at Ontotext wanted to demonstrate key company capabilities by building a tool to help track market performance of companies in various domains and identify the impact of key financial events and better understand global business trends and patterns. Ontotext took the opportunity to combine multiple databases, read and analyze unstructured text and thus draw conclusions from it, such as with news content serving information about market-moving events.

 

The goal

 

When mergers, acquisitions, global lawsuits, legal disputes or settlements are announced, these events are often reflected in the stock market performance of these companies. For example, a report on a pending merger between two companies could trigger a positive or negative influence on their position in the market. These movements have a material impact on how companies are perceived by the public, their shareholders and potential investors. Being able to detect and analyze these potential “Risk and Opportunity” events, as Ontotext calls it, is crucial for risk management, investment decisions, strategic planning and predictive analysis, as a few examples. 

 

While Ontotext’s GraphDB would serve as the underlying RDF database for storing and processing data, they desired an end-to-end solution that offered:

 

  • A platform to capture relations between concepts and relevant risk and opportunity events in a semantic model

  • An interface to easily detect pertinent events and analyze emerging trends and relations between events 

  • Model-driven extraction to be able to quickly analyze data from multiple sources and reconfigure for changing interests

  • An interface that allows customization of said model, depending on domain of interest

 

The challenges

 

One of the main challenges the Ontotext team faced when beginning this project was wanting to develop a prototype quickly with a lean team despite evolving internal requirements. The unit responsible for the solution had few developers to help build the back-end and web application and therefore required a tool that could get the platform in production quickly and efficiently.

 

Another challenge was that one of the main functionalities the team at Ontotext wanted their solution to have was the ability to edit and adapt their ‘events schema’, as they call it (we typically refer to this as an ontology or semantic knowledge model), without having to constantly rebuild the platform following changes to the model. They wanted customers to be able to customize and define what is considered an important event to them and be able to extract key information depending on their domain of interest. 

 

The solution: Ontotext RED

 

The final solution became Ontotext’s Relation and Event Detector (RED), a platform to identify and assess key business events. The platform combines a knowledge graph with large language models (LLMs), to offer search, exploration, analysis and sharing capabilities via an end-user interface application built on top of metaphactory. Ontotext RED leveraged metaphactory for its flexible and extendable semantic knowledge modeling to power its model-driven relation extraction and metaphactory’s intuitive low-code application building for its end-user interface.

 

Ontotext RED’s core superpower is its model-driven relation extraction. The underlying model is completely customizable and adaptable so that users can explicitly define which events are considered important to them, but also describe the parties involved, the relations between these actors, subsidiary companies, etc. The integration of LLMs allows for automation and scalability, relieving customers from the laborious process of traditional relation extraction. 

 

With the model, customers can easily identify and extract information from unstructured text, whether in the form of a news article or a passage in a report, that is essential to their organization. Integration of Wikidata adds additional richness by providing information such as the year a company was founded or the bio of the founder of a company involved. The additional context and semantic meaning enable organizations to gain a complete picture of a situation and make informed strategic decisions based on this information. Business analysts using the platform are then better equipped to analyze financial patterns, trends or even global business influence.

 

 

processed document with relations extracted and represented

Image: processed document with relations extracted and represented

 

Model-driven apps and model-driven extraction

 

Using a model-driven approach enables companies to transform their data and extend its value in numerous ways. Once data is captured in a knowledge graph and semantically modeled and linked, it can be used to build intuitive end-user interfaces, train LLMs for recommendation systems or surface insights for data analytics, all while following company-specific concepts, vocabularies, etc. 

 

A low-code, model-driven approach as used by metaphactory allows application engineers to derive the interface design and configuration directly from the ontology, like the configuration of forms to modify or enter data, search interfaces including filters and facets, and many more aspects of a modern user experience. This means that application engineers can focus on defining the interaction patterns for the data, select the right components to best visualize what is important, and minimize the effort in building and maintaining these user interfaces.

 

Additionally, because the user interface is not hard-coded but powered through the ontology, every time the data model or the data itself changes, the UI adapts automatically.

 

Using a low-code, model-driven approach you can build small, independent applications for a specific use case, to generic, company-wide data hubs or FAIR Data platforms. 

 

Since metaphactory follows FAIR data practices, it ensures that data is reusable, interoperable with multiple systems, and both human-understandable and machine-interpretable, making it possible to power such diverse use cases. 

 

To learn more about model-driven app building, read this blog post on “The composable enterprise”.

 

Low-code application building

 

metaphactory’s low-code platform was especially beneficial to Ontotext during this project considering that the unit responsible didn’t have developers to create their own applications—whether web, backend or otherwise. They decided to use metaphactory for its capabilities in offering quick model-driven development of the UI and direct access to the database for those purposes. 

 

This saved the team a significant amount of time and effort since metaphactory offers great startup setup and robust backend support. They were able to create an application to sit on top of metaphactory and utilize the knowledge graph built underneath. metaphactory’s low-code approach to app-building and parameterizable components means teams are able to produce a working application with low development efforts and in little time.

 

Supports lean teams & fast-track production

 

metaphactory was able to support the team responsible for this solution in producing a sophisticated knowledge graph platform for relation and event detection, despite having a lean team and evolving criteria. In only a few short months, Ontotext RED was ready for demos to multiple customers. When equipped with defined briefs and set requirements, the typical process of transforming an idea into a proof of concept in metaphactory is a mere 1-2 weeks. 

 

With metaphactory, we were able to reduce development time by 50%. This project would have taken approximately six months to complete, including back-end development and visual web application. - Antoniy Kunchev, Ontotext

 

Supports a repeatable stack

 

Ontotext was able to have its desired events schema, one that is extendible and customizable for easy reconfiguration for demos and out-of-the-box purchases for the customer. 


While the unit was initially unfamiliar with metaphactory, they were aware of its advanced capabilities and extendability. Quickly, they discovered the ease with which they could navigate the system and develop their knowledge graph application using metaphactory’s model-driven, low-code approach. This challenge proved to simply be a customary part of onboarding a new tool - and the subsequent building was ultimately an easy task.

 

Business benefits

 

  • Low-code application building

  • Semantic knowledge modeling

  • Ontology management

  • Follows FAIR open data principles

  • SPARQL connection

 

Why choose metaphacts & Ontotext

 

About metaphacts 

metaphacts helps global enterprises transform data into consumable, contextual and actionable knowledge. Our low-code, FAIR Data platform metaphactory simplifies capturing and organizing domain expertise in explicit semantic models, extracting insights from your data and sharing knowledge across the enterprise.

 

metaphactory includes features and tools for: 

 

  • Semantic knowledge modeling — explicitly capture knowledge & domain expertise in a semantic model & manage knowledge graph assets such as ontologies, vocabularies and data catalogs 

  • Low-code application building — build easy-to-configure applications that fit your enterprise and use-case requirements using a low-code, model-driven approach 

  • End-user-oriented interaction — users of any level of technical experience can interact with your data through a user-friendly interface that includes semantic search, visualization, discovery & exploration and authoring

 

Have a great idea for a new product or solution? With metaphactory, you can get from idea to proof-of-concept in only a couple of weeks and even fast-track to production in a matter of months. 

 

 

With Ontotext RED, organizations can now get ahead of potential market-influencing events and make strategic decisions informed by advanced insights and global trends. The events schema makes it easy to customize the model and identify risks and opportunities most relevant to your organization, including outcomes from other businesses that you can learn from. 

 

About Ontotext

 

Ontotext has been helping enterprises identify meaning across diverse datasets and massive amounts of unstructured information for over 20 years.

 

Ontotext specializes in knowledge management, understanding unstructured data, planning advanced initiatives such as data fabric, data mesh, Generative AI and Large Language Models (LLM) or a specific focus on fraud detection, drug formulary planning, Customer 360 and much more.

 

Combining a highly scalable and robust RDF database like GraphDB with metaphactory’s robust capabilities for semantic modeling and application-building created a unique and powerful platform that’s flexible and scalable. 

 

Does this resemble your particular business needs?


Contact us to speak with one of our experts and discuss your use case!