Esse artigo pode ser lido em:

You will at some point soon encounter a knowledge graph. It is also possible to create advanced applications and dashboards that generate high value for the organization, but to do so we need to expand data integration, providing data to those who need it, over time.

You will at some point soon encounter a knowledge graph, also known as a knowledge graph, and you will understand its power. To that end I conducted extensive research and wrote this text: to help you use this power to your advantage.

The extensive use of knowledge graphs allows questions posed by all types of users to be mapped onto an organized set of information that can provide the answers we seek.

And that is why they will be so fundamental in the very near future.

I prepared, together with Google’s NotebookLM, this audio summary of the content you have begun to read. It is a very interesting interpretation of how an Artificial Intelligence model understood this article of mine.

The Power of the Knowledge Graph

Have you ever used platforms such as Google Home, Siri, or Alexa? They are present on mobile phones and computers around the world, attempting to answer the questions that billions of people ask all day long.

These tools are equipped with different versions of knowledge graphs—a Knowledge Graph is an integrated collection of information about entities, which can contain a large number of links between the entities.

But what is a knowledge graph and what is its utility?

According to Wikipedia, a knowledge graph or Knowledge Graph in English is a knowledge base of Google’s search system that aims to improve the results of its search tool with information from semantic search.

With a graph, the model created to handle data can be as large, complex, and deep as you need it to be, because it deals with connections between complete, high-quality data, allowing the use of sparse and incomplete data, making them usable.

In summary, the knowledge graph uses semantic patterns to describe the structure of information to support reasoning and inference.

Graphs can have a predefined structure for node and edge properties, but their loading requires understanding which edges should be created and which nodes will relate to them.

But before we begin to discuss the use cases of a knowledge graph in your organization, we need to talk first about entities.

What is an entity?

Entities are the basic elements of a graph that can represent concepts, objects, things, or people. For those who remember Portuguese classes in school, we can compare them to nouns.

Therefore, this type of graph graphically represents the relationships and connections between various concepts. They allow us to infer various things, such as:

• Which album a song is part of and the relationship between the composers of each;
• How many clinical trials used a particular research work;
• How many projects a given professional is related to in your company.

The creation of a knowledge graph can be described as a semantic web, a network that knows things by connecting a whole range of information, allowing that same network to support a search.

Knowledge graphs are powerful for several reasons.

When they are based on semantic patterns, they allow us to relate knowledge to language in an orderly manner.

There are several reasons that justify the investment of time and effort to create a graph, among them the growing amount of information we need to handle in order to make decisions, the lack of standardization of this same data, impacting the quality of this same data and, ultimately, the insights drawn from this information.

To solve these problems we need to understand the structure that connects these entities, and the knowledge graph is the only currently implementable and sustainable way to do so.

Knowledge graphs for business analysis

Knowledge graphs are essential for many companies today: they provide the structured data and factual knowledge that guide many products and make them more “intelligent.”

In general, a knowledge graph describes objects of interest and connections between them.

For example, a graph can have nodes for a project in your company, the employees involved in that project, the director responsible for its execution, and so on. Each node can have properties such as the role of each employee, their knowledge and time in the company, for example.

There can be nodes for various projects involving a given employee. The area director can then traverse the knowledge graph to collect information about all the projects in which a given employee is involved, analyzing whether they are overloaded or not, inferring, for example, the reasons they appeared or, if applicable, were assigned.

Language models combined with knowledge graphs are the future of search

Bill Slawski, Dawn Anderson, Andrea Volpini, and Jason Barnard discuss the pros and cons of MUM and KELM and how they may affect the evolution of the Knowledge Graph.

More specifically, we will examine the relationship between language models, Google algorithms, and the Knowledge Map, and shed light on things we can do usefully to improve our entities in the Knowledge Map using language (thus, how language models and knowledge graphs interact).

Practical implementations of a graph in your business

Many practical implementations impose constraints on links in knowledge graphs, defining a schema or ontology.

For example, a link from a project to its director should connect an object of type Project to an object of type BusinessDevelopment. In some cases, the links themselves can have their own properties: a link connecting an employee and a project can have the name of the specific task that the employee developed.

Similarly, using an example from outside the world of business, a link connecting a politician to a specific role in government can have the time period during which the politician held that role.

Knowledge graphs and similar structures generally provide a shared substrate of knowledge within an organization, allowing different products and applications to use similar vocabulary and reuse definitions and descriptions created by others.

In addition, they often provide a compact formal representation that developers can use to infer new facts and build knowledge—for example, using the graph that connects projects and directors to discover which employees frequently work on projects together.

To solve your company’s information problems

Use knowledge graphs

But before leveraging the full power of a knowledge graph, it is necessary to recognize that there is a problem in companies:

Information is separated into silos.

The existence of multiple information silos or data silos (which are closely linked and can relate to one another) can be resolved with the strategy of creating graph layers.

It is possible to group all information about customers and purchases, collecting everything into a graph, connecting various groups to one another. Even while maintaining silos, we can group information into product categories and different levels of detail about the products in layers.

Use knowledge graphs in business

In a business environment, we can use graph structures due to such a necessary functionality in business analysis: connecting answers through links to related information.

But for a graph model to be functional within an organization, it is necessary that we can ask questions in simple ways and then be guided by a process that recommends and suggests new ways to continue asking questions.

This pattern of questions that generate answers connected to new questions is, in my view, the greatest advantage of using a knowledge graph in companies.

Knowledge graph as an exploratory tool

Creating an environment to explore information becomes useful when it is supported by advanced dashboards and applications that help us increase our productivity.

To that end, companies need to adopt an approach in which all their data can be integrated and interconnected into one large, coherent knowledge graph.

It is common to deal with projects composed of harmonized data originating from many distinct underlying systems, which continue to generate data uninterruptedly, hence the need to structure data using a graph.

Structuring this data is the key to dealing with the enormous volume of information, with document-oriented projects of all kinds.

From data lake to knowledge graph

This is a necessary step to make data truly useful for the organization, and if your company migrates to this model, it will be reaching a higher level of data integration.

Graph query languages can express complex queries that can filter information precisely. In addition, graphs can also be explored by algorithms and automated analyses.

If you build a graph correctly, it is possible to answer questions that would be practically impossible to answer if the data were in an SQL database.

But beyond that, if you create a graph with semantic patterns, the queries can be more powerful. You can use the information about the structure to automate queries, algorithms, and analyses and make them intelligent.

This automation depends on the co.com.br/o-que-sapan id=”urn:enhancement-26″ class=”textannotation”>o-dados-estruturados/” data-type=”post” data-id=”120″>correct markup of information along the path and the use of a id=”urn:enhancement-5ce5b3fa-41fd-41c4-8aae-ef18f94d8651″ class=”textannotation”>structure for the graph defined by an ontology.

Voice-based systems and graphs

One of the reasons voice-based and natural-language systems are so powerful is that you and I can ask a question and get the answer. Simple as that.

The knowledge graph is there, but under the hood. So when I receive an answer, the question passed through the structure of the graph, and the answer shows me its structure, represented in the complementary answers with links to related information.

The Power of the Knowledge Graph: use without restraint

The knowledge graph is the process for creating, loading, updating, and querying that must be scalable to prevent limitations, primarily technical ones, from determining the scope and reach of the graph.

With a graph project implemented in your organization, you will be able to identify the use cases that each knowledge graph addresses, the benefit that each will generate. With this understanding, it will be possible to create the graphs in less time and continuously evolve in their development.

This is extremely important for presenting users with a guided process for using the information described in the graphs we create, gradually expanding the scope of data integrated into a knowledge graph, connecting through links between the graphs, making them increasingly useful.

It is also possible to create advanced applications and dashboards that generate high value for the organization, but to do so we need to expand data integration, providing data to those who need it, over time.

Knowledge graphs simply help us do a better job.