?How do semantic partners differ from ML?
^^^add analytic animation here
ML triple stores enable & complement many diff analytics
Mach learn + NLP > entity extract > store in ML
Triples & inferencing > create knowledge graphs & support data mining & analytics
Let’s talk just a bit more about terminology as the semantics of semantics can get messy…
From our perspective, MarkLogic is a document oriented database. It also has a built-in triple store. It also serves as a precursor to other capabilities.
A triple store can serve as a knowledge graph with facts and relationships about billions of things.
You can also mine data in MarkLogic to do predictive analytics.
MarkLogic also works with partners such as SmartLogic to do entity extraction. When you take free text and extract facts about people, places, and things, the unstructured data becomes machine readable.
This is the basis for natural language processing, or NLP.
NLP, machine learning, and pattern recognition fall into the realm of cognitive computing, and all of this is encroaching into the territory of AI.
You guys recognize these logos?
OF COURSE! Everybody knows “ENTERTAINMENT COMPANY”
Actually, the entertainment companies that are our customers, and those we hope will honor us by becoming customers,
…are doing or looking to do similar things with semantics, albeit in their own way…so we talk about them as a group in this presentation.
As David Gorbet said in his keynote, we’re helping enterprises
See their entire product and customers through intelligence search
Understand their production and distribution processes through a Semantic Metadata Hub
Deliver customized, targeted content and user experience through Dynamic Semantic Publishing
Maintain a valuable engagement with customers through semantically driven Recommendations
AND assess business risks by leveraging semantics for Compliance
The goal of this first session is to de-mystify "Semantics".
Some people are put off by the notion that semantics is somehow magical, or at least enormously complicated.
Semantic technologies can be very powerful, but there's no magic here – just science.
By a show of fingers – where 0 is "I don’t know what a triple is" and 10 is "I have a PhD in Ontologies" – how much do you know about Semantic technologies?
I see some 7s and 8s – OK, I'll go through this section quite quickly.
For the 1s and 2s, I'll define some basic terms and give you a general idea of what we mean by Semantics in this context.
BBC – DSP
BSI – Semantic Search
InfoBox
There’s a lot you can do with JUST triples, but the real magic of semantics comes when you use triples alongside documents.
Here, you can see a document as Marklogic sees it. A document is stored as XML or JSON, which is a hierarchical tree format.
Documents are schema-agnostic, human-readable, and don’t carry all of the entity integrity constraints that you had with a relational model.
You can do a lot with documents, but even documents can fall short when it comes to optimizing for facts and relationships, which are best stored in a graph model as triples.
Here, you can see a graph formed by triples about a particular video title.
In this graph we know that a title not only has this metadata, but the title has characters, etc.
With a single query, you can bring back the document, parts of the graph, or both and have it all materialize at runtime.
This “multi-model” view of data provides more flexibility and agility than any other model.
If you want to go into more detail about how semantics helps with classification
Mix'n'Match documents and triples
There’s a lot you can do with JUST triples, but the real magic of semantics comes when you use triples alongside documents.
Here, you can see a document as Marklogic sees it. A document is stored as XML or JSON, which is a hierarchical tree format.
Documents are schema-agnostic, human-readable, and don’t carry all of the entity integrity constraints that you had with a relational model.
You can do a lot with documents, but even documents can fall short when it comes to optimizing for facts and relationships, which are best stored in a graph model as triples.
Here, you can see a graph formed by triples about a particular video title.
In this graph we know that a title not only has this metadata, but the title has characters, etc.
With a single query, you can bring back the document, parts of the graph, or both and have it all materialize at runtime.
This “multi-model” view of data provides more flexibility and agility than any other model.
If you want to go into more detail about how semantics helps with classification
There’s a lot you can do with JUST triples, but the real magic of semantics comes when you use triples alongside documents.
Here, you can see a document as Marklogic sees it. A document is stored as XML or JSON, which is a hierarchical tree format.
Documents are schema-agnostic, human-readable, and don’t carry all of the entity integrity constraints that you had with a relational model.
You can do a lot with documents, but even documents can fall short when it comes to optimizing for facts and relationships, which are best stored in a graph model as triples.
Here, you can see a graph formed by triples about a particular video title.
In this graph we know that a title not only has this metadata, but the title has characters, etc.
With a single query, you can bring back the document, parts of the graph, or both and have it all materialize at runtime.
This “multi-model” view of data provides more flexibility and agility than any other model.
If you want to go into more detail about how semantics helps with classification
There’s a lot you can do with JUST triples, but the real magic of semantics comes when you use triples alongside documents.
Here, you can see a document as Marklogic sees it. A document is stored as XML or JSON, which is a hierarchical tree format.
Documents are schema-agnostic, human-readable, and don’t carry all of the entity integrity constraints that you had with a relational model.
You can do a lot with documents, but even documents can fall short when it comes to optimizing for facts and relationships, which are best stored in a graph model as triples.
Here, you can see a graph formed by triples about a particular video title.
In this graph we know that a title not only has this metadata, but the title has characters, etc.
With a single query, you can bring back the document, parts of the graph, or both and have it all materialize at runtime.
This “multi-model” view of data provides more flexibility and agility than any other model.
If you want to go into more detail about how semantics helps with classification
There’s a lot you can do with JUST triples, but the real magic of semantics comes when you use triples alongside documents.
Here, you can see a document as Marklogic sees it. A document is stored as XML or JSON, which is a hierarchical tree format.
Documents are schema-agnostic, human-readable, and don’t carry all of the entity integrity constraints that you had with a relational model.
You can do a lot with documents, but even documents can fall short when it comes to optimizing for facts and relationships, which are best stored in a graph model as triples.
Here, you can see a graph formed by triples about a particular video title.
In this graph we know that a title not only has this metadata, but the title has characters, etc.
With a single query, you can bring back the document, parts of the graph, or both and have it all materialize at runtime.
This “multi-model” view of data provides more flexibility and agility than any other model.
If you want to go into more detail about how semantics helps with classification
We can think of "Order", "Order associated with App1" as metadata; "extended" because we can extend the link from App1 to "application that requires TopSecret".
International Classification of Diseases, a set of codes used by physicians, hospitals, and allied health workers to indicate diagnosis for all patient encounters.
International Classification of Diseases, a set of codes used by physicians, hospitals, and allied health workers to indicate diagnosis for all patient encounters.
The goal of this first session is to de-mystify "Semantics".
Some people are put off by the notion that semantics is somehow magical, or at least enormously complicated.
Semantic technologies can be very powerful, but there's no magic here – just science.
By a show of fingers – where 0 is "I don’t know what a triple is" and 10 is "I have a PhD in Ontologies" – how much do you know about Semantic technologies?
I see some 7s and 8s – OK, I'll go through this section quite quickly.
For the 1s and 2s, I'll define some basic terms and give you a general idea of what we mean by Semantics in this context.