Challenge

Storing information in knowledge graphs and ontologies can make it much easier to understand the relationships between objects, processes and actors in your business and to extract insights that can make your business more efficient.

Our client sells a large range of electronic consumer goods and wanted to build an ontology to represent these goods that can be interrogated at a granular level to facilitate customer comparisons and conversational queries about particular product features. They are a multinational company with federated business units and siloed data models and needed a flexible way to interrogate different features and products.

We were asked to build a prototype knowledge store mapping facility that allows the user to import product information from a range of sources including SOAP and REST apis into the knowledge store.

 

Solution

Filament built a data mapping platform that allowed the user to import data from APIs by establishing a mapping between the incoming data format and the knowledge graph fields in a graphical user interface.

Much of the data was accessible via APIs but due to the siloed nature of our client’s business, some of the data could not be accessed via backend systems. We therefore developed a website scraping approach that could be used to map information displayed on parts of the client’s public website into their knowledge graph framework. This approach provided a high degree of flexibility, allowing the client to map information displayed on any HTML-based web page into their knowledge graph with ease.

The POC project was carried out over 12 weeks.