EKGF Community Voices
EKGF Community Voices articles are written by members of the Enterprise Knowledge Graph Foundation. All opinions are the properties of the authors and do not necessarily reflect the views of Enterprise Knowledge Graph Foundation.
Enhancing Customer Experience Through Airbnb's Knowledge Graph
by Carlos Tubbax
In order to make the utility of knowledge graphs for enterprises clearer, it is important to provide some examples of success stories in the adoption of knowledge graphs by businesses.
People using Amazon’s Alexa, ordering food through Uber Eats, or booking a listing on Airbnb are using a knowledge graph even if they are totally unaware of that. Amazon, Airbnb, Uber among other big players have been using knowledge graph technology to disrupt entire sectors and to create new value propositions for their customers.
For example, Amazon uses a product graph to categorize products on its retail website and to make better product suggestions to customers, among other things. In that context, knowledge graphs fulfill the role of Disruption Enabler. In order to illustrate this, this paper will discuss how Airbnb uses a knowledge graph to enhance its customer experience by integrating and conducting computer reasoning on data coming from different data sources.
Enterprise Knowledge Graphs may also provide great value when being used to support the daily operations of a business and, in that context, the enterprise is their end user. This role is known as the Supporting Role. Although there are fewer examples of knowledge graphs being used in this context, Deloitte’s white paper  discusses how knowledge graphs could be used in organizations to unravel the intricacies of their own business processes, relationships, supply chains, etc. as a business supporting/enabling technology.
Customer Experience Enhancement at Airbnb
In order to move towards its vision of becoming an end-to-end travel platform, Airbnb needs to be able to provide customers with insights that help them decide when to travel, where to travel and what to do on their trips. For instance, Airbnb needs to be able to answer queries such as:
“What are the most popular landmarks and neighborhoods in London?”
“Which Airbnb listings are best suited for working nomads?”
“What are the most popular Italian restaurants in New York?”
Answering these queries may help travelers plan their trips better and, in turn, Airbnb may increase its customer value proposition.
As a means to answer all these customer queries, Airbnb uses a knowledge graph represented in Figure 1. As explained by Chang (2018), semantic web knowledge graphs offer the ability of structuring and adding meaning to data from different sources, such as relational databases, in a scalable way to answer these queries.
In this way, Airbnb is moving forward into becoming an end-to-end travel platform that serves its customers throughout their entire journeys instead of only renting listings out.
Risk Management with Knowledge Graphs According to Deloitte
Deloitte defines a knowledge graph as; “a means to connect and represent knowledge in an area of interest using a graph-like structure. It is typically built on top of existing databases to link data together at web-scale, combining both structured and/or unstructured information.. In contrast to relational databases, graph models offer the ability to link concepts or entities to one another and to represent these relationships. On top of that, human expertise can be added to provide context to data as machine-readable definitions embedded in the knowledge graph. According to Deloitte, knowledge graphs might help businesses in two related areas as follows:
As companies rely steadily more on Artificial Intelligence, context will be more necessary than ever to provide meaningful data to these AI-driven applications as these models are only as good as the data they feed on. This context could be added to the equation with knowledge graphs that incorporate meaning, defined by human expertise, within data. This may result in better informed AI models for decision making.
The capacity to connect datasets in a meaningful way is strategic for data-driven businesses as it enables decision makers, users, and also AI applications, to make better decisions. Additionally, connecting data spread across a company may help it to understand possible threats, opportunities, and themselves better.
Deloitte provides three use cases to illustrate this:
360° View of Risk and Value. As companies are overwhelmed with increasing amounts of data, they need to integrate and harmonize these data based on their meaning. Knowledge graphs are enablers to semantically integrate diverse data and draw connections at a large scale. Knowledge graphs allow users to connect different sources of data in a scalable and efficient way regardless of their underlying formats and models. For instance, an investment bank that needs to conduct due diligence on its potential customers to avoid sanctions could use a knowledge graph to integrate data about family relationships, financial transactions, and regulations so that it makes a better risk assessment of its customers.
Compliance Management. As businesses need to comply with evolving and increasing regulations, policies, and contracts, compliance is a key challenge. Knowledge graphs offer the capacity to not only unify and interlink different sources of compliance data but also to conduct complex automated compliance checks through computer reasoning. For instance, a company that wants to enter a new market could use a knowledge graph to integrate the different compliance data sources of that country, to identify the relevant regulations, and to set automated compliance rules.
Data Lineage and Metadata Management. Tracking data throughout their lifecycle is daunting but also a mandatory task for companies as this data needs to be accessed, utilized, transformed, merged while having to comply with certain regulations such as GDPR. The ability of knowledge graphs to integrate and harmonize heterogeneous data sources gives a business the opportunity to have a layer that provides a full view of the data in the company and its lifecycles. For instance, a company in a merger might be able to get an overview of its own data throughout its lifecycle together with the sources used and all their dependencies to make better decisions for data integration and management.