Learning Library

← Back to Library

Understanding Knowledge Graphs and Their Uses

Key Points

  • Knowledge graphs power virtual assistants by storing semantic relationships—e.g., “Ottawa” linked to “Canada” via a “capital” edge—enabling quick factual answers.
  • They are composed of nodes (entities) and edges (relationships), allowing multiple, diverse connections between the same entities (e.g., Paris → France as “capital” and Paris → Roman Empire as “city of”).
  • By integrating multiple data sources (census data, online reviews, etc.) into a unified graph, statistical inference can fill gaps and reveal more accurate insights, such as estimating the true number of Chinese restaurants in New York City.
  • Natural language processing is used for semantic enrichment, converting unstructured text into linked nodes and edges, which supports not only question‑answering but a wide range of commercial applications.

Full Transcript

# Understanding Knowledge Graphs and Their Uses **Source:** [https://www.youtube.com/watch?v=y7sXDpffzQQ](https://www.youtube.com/watch?v=y7sXDpffzQQ) **Duration:** 00:05:34 ## Summary - Knowledge graphs power virtual assistants by storing semantic relationships—e.g., “Ottawa” linked to “Canada” via a “capital” edge—enabling quick factual answers. - They are composed of nodes (entities) and edges (relationships), allowing multiple, diverse connections between the same entities (e.g., Paris → France as “capital” and Paris → Roman Empire as “city of”). - By integrating multiple data sources (census data, online reviews, etc.) into a unified graph, statistical inference can fill gaps and reveal more accurate insights, such as estimating the true number of Chinese restaurants in New York City. - Natural language processing is used for semantic enrichment, converting unstructured text into linked nodes and edges, which supports not only question‑answering but a wide range of commercial applications. ## Sections - [00:00:00](https://www.youtube.com/watch?v=y7sXDpffzQQ&t=0s) **Explaining Knowledge Graph Basics** - The speaker introduces knowledge graphs, showing how they represent entities and multiple relationships—such as capitals, historical affiliations, movies, or recipes—to enable machines like virtual assistants to retrieve and connect contextual information. - [00:03:11](https://www.youtube.com/watch?v=y7sXDpffzQQ&t=191s) **Semantic Enrichment via Knowledge Graphs** - The speaker explains how natural language processing creates semantically enriched knowledge graphs from unstructured text and highlights varied commercial uses—including video recommendations, insurance fraud detection, and retail product pairing—illustrated with a simple human‑coffee‑sleep graph. ## Full Transcript
0:00Perhaps you're unfamiliar 0:02with the term of 0:04knowledge graph, 0:06but I suspect you 0:08have benefited from one, maybe 0:11even today. 0:13Take your favorite virtual 0:15assistant. 0:16Did you know that when you ask 0:18a question like what is the capital 0:20of Canada? 0:22The assistant is pulling the 0:24response. 0:24Ottawa in this case from information 0:26in a knowledge graph 0:29and not as graphs can be seen as a 0:31way of representing semantic 0:32information between two entities. 0:34And what's really cool about them 0:36is that modern applications allow 0:38almost any entity you could imagine 0:40to be described with anyone. 0:41For example, we could have a 0:42knowledge graph of movies and 0:44actors, or we can describe 0:46ingredients or recipes, as well as 0:48the steps required to cook them. 0:49And this means machines are able to 0:51understand how these entities 0:53relate to each other, along 0:55with a shared attributes, and this 0:57allows us to draw connections 0:59between different things in the 1:00world around us. 1:02Now, a knowledge graph is made 1:04up of nodes, 1:08and connected by 1:10edges. 1:12Nodes describe any object 1:14or person or place, and an 1:16edge defines the relationship 1:18between the nodes. 1:20So, for example, a 1:22node of Ottawa.. 1:23and 1:27a node of Canada.. 1:32might be connected by the edge 1:35of capital. 1:40And the thing here is that 1:42the pair of nodes, they can be 1:43connected by more than one relation. 1:46If the two are related in multiple 1:48ways. So for example, 1:50let's take another city. 1:52Let's take Paris. 1:53Now, Paris 1:56is the capital of 1:58France, 2:03but it is also part 2:05of the Roman Empire. 2:08Or it did used to be. 2:13So in this case, the edges are 2:15Paris to France is capital. 2:19And then Paris to Roman 2:21Empire is 2:23city of. 2:26And we can see then how that these 2:28nodes can be connected with multiple 2:30edges as we expand this. 2:32And knowledge graphs can sort of 2:34build different data sources 2:36and bind them together to infer 2:38missing facts. So let's say you're 2:39trying to predict the number of 2:41Chinese restaurants in New York 2:42City. 2:43You could use one data source, let's 2:44say census data, but that might not 2:46tell you the whole story. 2:47It might be out of date. 2:49It might not classify everything 2:50correctly and so forth. 2:52So if we had a second data source 2:53like, say, online reviews about 2:55all the different restaurants and 2:57put them all in a knowledge graph, 2:59then we can use statistical methods 3:01to infer that actually there are 3:03two thousand nine hundred 3:04restaurants serving Chinese food 3:06in New York City, which may be a lot 3:08different than what was reported in 3:09the census data alone. 3:11Now, knowledge graphs utilize 3:14something called natural 3:17language processing. 3:22Or abbreviated 3:24to NLP 3:26to construct a view of 3:28nodes and edges through a process 3:30called semantic enrichment. 3:31I can take some unstructured text, 3:33say a white paper, and 3:35classify that text using natural 3:38language processing 3:40to really sort of create datasets 3:43which are correlated and 3:45related to that 3:47information. 3:49And one that builds me is a 3:50knowledge graph. 3:52And beyond sort of helping with 3:54question and answer queries, there 3:55are a lot of other commercial 3:56applications for knowledge graphs. 3:59So, for example, the recommended 4:01videos that are probably appearing 4:02alongside this one in YouTube right 4:04now. Well, they leverage 4:06a knowledge graph based on queries 4:08people are searching for and other 4:09videos that they might enjoy. 4:11In insurance you can use knowledge 4:13graphs to sort of and make sure 4:15that a given claim 4:17for damage is actually 4:19a true claim, or whether it's been 4:21one that's reported by a 4:22policyholder for fraud. 4:24And in retail, knowledge graphs 4:26can assist understanding the 4:27relationship between products so 4:29that companies can recommend 4:30different pairings that might be of 4:31interest to customers. 4:34And I'll leave you here with some 4:35wisdom that I was acutely aware 4:37of last night and I'll 4:39share in knowledge graph form. 4:42It consists of three 4:44nodes. There's human. 4:49Then there is coffee. 4:54And then there is sleep. 4:57And these note connected, of 4:59course, by edges. 5:01The edge between human 5:03and coffee is consumed. 5:07The edge between human 5:09and sleep is needs. 5:13And what I learned last night, the 5:15edge between coffee and sleep, 5:18is prevents. 5:22Avoid caffeine after five p.m. 5:24folks. 5:25If you have any questions, please 5:27drop us a line below, and if 5:29you want to see more videos like 5:30this in the future, please like and 5:32subscribe. 5:33Thanks for watching.