Management consulting firms are doing their best to help clients navigate the age of AI. To stay ahead of the competition and ensure their clients’ success, they should encourage companies to build a knowledge graph and improve data governance.
The Challenge Of Big Data
Big data has three characteristics that put companies to the test: variety, volume and velocity. Known as “the three Vs of big data,” they describe how data often exists in different types and large volumes, in addition to being created at large speed.
If that weren’t enough, a lot of relevant business information is also stored in “unstructured data” — the industry jargon for text. However, human language is often too complex for machines to understand. There are techniques which try to tackle this problem, but most of them cannot fully grasp the meaning and correctly interpret the nuances of language.
Consulting firms have stepped in to help their clients extract insights from big data. However, they are having a hard time fulfilling their promise due to their over-reliance on machine learning algorithms. That is largely due to their clients’ mismanagement of data.
The Struggle To Improve Data Quality
Poor data quality keeps companies from gaining timely and accurate insights from data that is essential for business decisions. In fact, an MIT article suggests that companies lose 15% to 25% of their revenue because of bad data quality.
Consequently, data analysis and presentation make up a significant part of today’s management consulting services. Consultants must go through the rather time-consuming task of gathering and cleaning data from disparate parts of their clients’ organizations. Only then are they able to process and interpret the data in a meaningful way. As a result, data scientists spend more than half of their time collecting and processing data before it can be used.
In an attempt to improve the situation, consulting firms rely even more on a technology they already know: machine learning. They deploy algorithms that help automate and speed up some of the data cleaning process. However, this is not a long-term solution. What they must do is help clients change how they handle data.
Adding A Knowledge Graph To The Mix
A knowledge graph is a representation of knowledge that acts as a mediator between humans and machines. It represents the world in ways very similar to humans, by linking ideas, concepts and things.
With a knowledge graph in hand, companies can solve many of their data problems. They can use it to automatically classify, link, validate and enrich all their data. The knowledge graph can also be used to connect different datasets and files to make them available as they were only one database. This enables companies to extract more value from data.
Management consulting firms could play a major role in supporting companies to build their enterprise knowledge graph. The process requires effort and collaboration from different stakeholders within a company, from executives to subject matter experts. This may sound like a lengthy and complex process, but it doesn’t have to be. There are methodologies, standards and tools that speed things up, such as automatically extracting thousands of concepts from texts and databases, for example.
This is essential to successfully implement AI strategies. A knowledge graph can even eliminate the need for large amounts of data to train machine learning algorithms (known as the “cold start problem”). When data is classified and linked, algorithms require less data to produce relevant results because they can analyze not only the data itself but also its context.
Another advantage of AI solutions built using a knowledge graph is that they are reusable. Once the knowledge is in place, it no longer depends on existing databases and becomes a standalone knowledge asset. This allows the knowledge graph to cover many areas a company works in to be used repeatedly.
Knowledge graphs also make machine learning algorithms more explainable. The mapping of relationships between data makes it easier for humans to understand how algorithms reach specific results. Decision-makers are then able to evaluate and trust the performance of their AI applications.
How To Help Clients Get Started With Knowledge Graphs
The best way to approach knowledge graphs is to start small and grow. Help your clients pick a concrete use case that can quickly show the value the knowledge graph can bring to their organization. Take a look at their strategic goals, select a use case that has well-defined business value and that makes process or services more efficient and intelligent. Once you have helped implement one use case, move to other ones until the organization completely adopts knowledge graphs.
Good use cases to get started include improving search capabilities, developing better recommendation systems or automating document classification. They are all low-hanging fruits of knowledge graph implementation. Once these are in place, you can encourage your clients to build their own AI applications such as chatbots — virtual assistants that extract insights from large amounts of unstructured text documents. Because they use a knowledge graph, these applications will be explainable and clients will be able to trust and understand their results.
The combination of knowledge graphs and machine learning is revolutionizing data analytics. This approach has many applications, from human resources to finance. Companies adopting it will extract more value from data and boost their innovation capabilities. It will become a standard in every industry, and management consulting firms should embrace it as soon as possible to stay relevant.
Sourced from Forbes - contributed by Andreas Blumaeur